Date   
Re: Was: Update on BITX20 Kit Progress - tell what the problem is please?

Arv Evans
 

Farhan

Thanks for your explanation and suggestions. That should help answer
some of the questions, and could provide direction for upgrade of
existing and future BITX transceivers.

Perhaps I should add a few more comments & clarifications about the new
kit PCB. It does include added gain and better passband shaping in a
new two transistor receive AF preamp. Moderately increased gain in the
receiver 2nd IF stage does maintain similar impedance characteristics as
the original design, thus potential problem with crystal filter
characteristics should be minimal. The overall look and feel of the
original design is still there.

In building Version 1.1 kit prototypes both Jim Kortge and myself are
doing tests that may have never been performed before, or at least not
documented and published. Jim is building his prototype using 11.0 MHz
crystals while my version uses the original 10.0 MHz IF frequency. The
change in IF frequency may be necessary for US based hams who live close
enough to WWVB 10.0 MHz transmitters for that to be a problem. We are
both learning some interesting things about the rig and about this
particular configuration. This information can be shared with the group
once we are finished with these prototype builds. Hopefully, this data
will then be used as a basis for continued improvement as the design
evolves over time.

As Farhan stated in his posting, there is room for improvement, and he
has recommended some of the possible changes.

Arv
_._



farhan@... wrote:



low audio:

the low audio output that some have complained about is traced to the
ineffective audio pre-amp. It is best to fix it there rather than
increasing the IF gain. I have mentioned this several times in my previous
posts. I think this needs a little more explanation. here goes ...

the receiver chain has just enough gain to overcome the losses in the
filters and maintain the low noise figure. each stage has an effective
gain of 16db. there is a total gain of 48db until the audio pre-amp
(Q1, Q2, Q3). the losses are mostly in the RF filter, diode mixers (two)
and the crystal filter. These are are correspondingly -3db, -7db, -7db
and -7db = -25db. hence, we still have an overall positive gain of
48-22 = 26db. hence, the problem is of low overall gain rather than
of the circuit being 'noisy' uptil the audio pre-amp. the culprit is low
gain out of Q4 and thats what needs to be fixed.

it is a simple modification to replace the single transistor audio amp
with a better, higher gain, lower noise figure amplifier using two
audio transistors. I have tried and recommend using the audio preamp
used in W7EL's 'optimized transceiver'.

while adding a two-stage IF amp also cures the problem, it introduces
a different problem. First, the series 220 ohms resistor feeding into the
base of the IF amplifier adds noise. second, the termination looking into
the crystal filter has changed. hence, the received and the transmitted
pass-band characterisitics of the crystal filter are different. this means
that though you may hear other side properly, there is no gaurentee that
the other side hears you equally well because the crystal filter sees a
different termination on transmit and receive!

here is what i suggest:
http://www.qrp.pops.net/images/popDC_a.gif
<http://www.qrp.pops.net/images/popDC_a.gif>

just use the circuit upto q1 and q2 as the audio pre-amp in place of Q4
of the original bitx20 design.

pa instability:

good construction practices will always reduce the chances of instability
and however stable your design maybe, you can always lay it out so that
it oscillates.

the power chain of bitx20 is a trade-off (like any engineering decision).
I couldnt afford to use RFCs as they are simply not available
off-the-shelf to many of us (i have none in my junkbox). Hence, the
currents (which are sizable in the PA chain) are not entirely decoupled
from power supply.
this is the main cause of instability in the power chain. On OM Wes's
suggestion,I fashioned RFCs by adding 20 turns to the TV baluns and
putting them
in the power lines of each of the transistors of the power chain. This
makes the power amplifier very stable.

The power amplifier should be ideally assembed the ugly way or the
manhattan way (to provide a low impedance return to ground) and it should
be mounted away from the main board (especially far away from the VFO to
avoid thermal drift).

As the power level on the verge of being QRO (6 watts), this project
needs
to treat its power amp just like any QRO power amp. with proper sheilding
between the power stage and the remaining transceiver circuitry. We
are no
longer in the 1 watt territory. nor is it a class-c which is inherently
more stable.

the final issue with a bidirectional transceiver is that the power
amplifier's input is always close to the output because the pa has to be
switched in and out of the signal path on transmit/receive. hence, using
two relays is recommended: one at the input of the power amp and one at
the antenna.

hence, the take away is:

a) power amp should be mounted away from the mainboard preferably
sheilded
as well(a good practice is to mount the two boards on either side of the
chassis).

b) use a lot of ground around the power amps.

c) put RFCs in the power line going to Q14 as well as the driver.

d) isolate the input and output of the power chain and layout the power
chainin a straight line.

- farhan

On Thu, 15 Mar 2007, Jim Strohm wrote:


--- arv evans <arvid.evans@...
<mailto:arvid.evans%40gmail.com>> wrote:

Hi

There are several mentions in historical posts
regarding instability and
low power output from the BITX20 design. The
original Farhan design
used a separate board (or more open spacing) for the
the RF PA section,
which seems to have been easier to tame than those
(like my personal
first construction) that had the PA section close to
other stages. Some
of the "inherently stable" designs become less than
stable if compressed
into small spaces where unwanted coupling becomes
problematic. The
problem is not in the original design. It is with
compressing the whole
rig onto a very tightly packed PCB layout with no
shielding between
stages.
Thank you, Arv.

In addition, after the first pass of the PCB layout,
we fixed a couple of inevitable early-design glitches,
and also adjusted some component values and placements
to help give a better final product at a lower final
cost.

This kind of tweaking in development is normal in the
real world. The difference from the real word for
BITX is -- with the exception of your humble tech
writer (that would be me), everybody else pretty much
has a regular fulltime job. So all the designers and
the beta builders have to dovetail their BITX
development around their daytime jobs.

My occupation right now is to find a good-paying gig
that supports my next political campaign (2008 season)
and also allows me to pursue work that fosters my
career as a free-lance writer, which includes helping
out on projects like this. So you know -- my "pay" to
help with this project is to help with this project.
Like everybody else except Doug, who has a significant
personal financial risk in this project, I'm working
for free.

So I hope everybody will understand why things have
not been progressing at the normal 40 hours a weel
schedule of industry.

We WILL deliver a product that will delight you. At
least that's been our first and only goal the whole
time.

Jim N6OTQ



__________________________________________________________
Expecting? Get great news right away with email Auto-Check.
Try the Yahoo! Mail Beta.
http://advision.webevents.yahoo.com/mailbeta/newmail_tools.html
<http://advision.webevents.yahoo.com/mailbeta/newmail_tools.html>

Ver.3 PCB mistake

ewcheeguan
 

Hi all,

I'm building my 2nd set of Bitx for my friend, and using the V3
single PCB layout. I found a mistake with the value of bias resistor
in between C and B of Q3. The schematic shows 4.7K which is seems
correct value but the PCB shows 1k .

Thanks.

Regards.
9W2ECG

BITX shoes -- more

Jim Strohm <n6otq@...>
 

A pass-along on the amp described in my BITX shoes
post. This is a description of the input and output
transformers.

Jim N6OTQ


The spec. on the ferrite's is anybody's guess but
the output
transformer secondary (to the antenna) is 3 turns of
thin pvc hook-up
wire on two binocular cores. The primary is a single
turn centre
tapped formed out of two sections of brass tube with
secondary
winding running through them. The input transformer
is a single, half
size, binocular core with the primary using 2 turns
and the secondary
3 turns, both in enamelled copper wire about 18 sw
gauge.

Best 73
G6AKG



____________________________________________________________________________________
The fish are biting.
Get more visitors on your site using Yahoo! Search Marketing.
http://searchmarketing.yahoo.com/arp/sponsoredsearch_v2.php

Re: Was: Update on BITX20 Kit Progress - tell what the problem is please?

farhan@...
 

low audio:

the low audio output that some have complained about is traced to the
ineffective audio pre-amp. It is best to fix it there rather than
increasing the IF gain. I have mentioned this several times in my previous
posts. I think this needs a little more explanation. here goes ...

the receiver chain has just enough gain to overcome the losses in the
filters and maintain the low noise figure. each stage has an effective
gain of 16db. there is a total gain of 48db until the audio pre-amp
(Q1, Q2, Q3). the losses are mostly in the RF filter, diode mixers (two)
and the crystal filter. These are are correspondingly -3db, -7db, -7db
and -7db = -25db. hence, we still have an overall positive gain of
48-22 = 26db. hence, the problem is of low overall gain rather than
of the circuit being 'noisy' uptil the audio pre-amp. the culprit is low
gain out of Q4 and thats what needs to be fixed.

it is a simple modification to replace the single transistor audio amp
with a better, higher gain, lower noise figure amplifier using two
audio transistors. I have tried and recommend using the audio preamp
used in W7EL's 'optimized transceiver'.

while adding a two-stage IF amp also cures the problem, it introduces
a different problem. First, the series 220 ohms resistor feeding into the
base of the IF amplifier adds noise. second, the termination looking into
the crystal filter has changed. hence, the received and the transmitted
pass-band characterisitics of the crystal filter are different. this means
that though you may hear other side properly, there is no gaurentee that
the other side hears you equally well because the crystal filter sees a
different termination on transmit and receive!

here is what i suggest:
http://www.qrp.pops.net/images/popDC_a.gif

just use the circuit upto q1 and q2 as the audio pre-amp in place of Q4
of the original bitx20 design.


pa instability:

good construction practices will always reduce the chances of instability
and however stable your design maybe, you can always lay it out so that
it oscillates.

the power chain of bitx20 is a trade-off (like any engineering decision).
I couldnt afford to use RFCs as they are simply not available off-the-shelf to many of us (i have none in my junkbox). Hence, the currents (which are sizable in the PA chain) are not entirely decoupled from power supply. this is the main cause of instability in the power chain. On OM Wes's suggestion,I fashioned RFCs by adding 20 turns to the TV baluns and putting them
in the power lines of each of the transistors of the power chain. This makes the power amplifier very stable.

The power amplifier should be ideally assembed the ugly way or the manhattan way (to provide a low impedance return to ground) and it should be mounted away from the main board (especially far away from the VFO to avoid thermal drift).

As the power level on the verge of being QRO (6 watts), this project needs to treat its power amp just like any QRO power amp. with proper sheilding between the power stage and the remaining transceiver circuitry. We are no longer in the 1 watt territory. nor is it a class-c which is inherently more stable.

the final issue with a bidirectional transceiver is that the power amplifier's input is always close to the output because the pa has to be switched in and out of the signal path on transmit/receive. hence, using two relays is recommended: one at the input of the power amp and one at the antenna.

hence, the take away is:

a) power amp should be mounted away from the mainboard preferably sheilded as well(a good practice is to mount the two boards on either side of the chassis).

b) use a lot of ground around the power amps.

c) put RFCs in the power line going to Q14 as well as the driver.

d) isolate the input and output of the power chain and layout the power chainin a straight line.

- farhan

On Thu, 15 Mar 2007, Jim Strohm wrote:


--- arv evans <arvid.evans@...> wrote:

Hi

There are several mentions in historical posts
regarding instability and
low power output from the BITX20 design. The
original Farhan design
used a separate board (or more open spacing) for the
the RF PA section,
which seems to have been easier to tame than those
(like my personal
first construction) that had the PA section close to
other stages. Some
of the "inherently stable" designs become less than
stable if compressed
into small spaces where unwanted coupling becomes
problematic. The
problem is not in the original design. It is with
compressing the whole
rig onto a very tightly packed PCB layout with no
shielding between
stages.
Thank you, Arv.

In addition, after the first pass of the PCB layout,
we fixed a couple of inevitable early-design glitches,
and also adjusted some component values and placements
to help give a better final product at a lower final
cost.

This kind of tweaking in development is normal in the
real world. The difference from the real word for
BITX is -- with the exception of your humble tech
writer (that would be me), everybody else pretty much
has a regular fulltime job. So all the designers and
the beta builders have to dovetail their BITX
development around their daytime jobs.

My occupation right now is to find a good-paying gig
that supports my next political campaign (2008 season)
and also allows me to pursue work that fosters my
career as a free-lance writer, which includes helping
out on projects like this. So you know -- my "pay" to
help with this project is to help with this project.
Like everybody else except Doug, who has a significant
personal financial risk in this project, I'm working
for free.

So I hope everybody will understand why things have
not been progressing at the normal 40 hours a weel
schedule of industry.

We WILL deliver a product that will delight you. At
least that's been our first and only goal the whole
time.

Jim N6OTQ



____________________________________________________________________________________
Expecting? Get great news right away with email Auto-Check.
Try the Yahoo! Mail Beta.
http://advision.webevents.yahoo.com/mailbeta/newmail_tools.html

Re: Question for Arv

Allison Parent <kb1gmx@...>
 

--- In BITX20@..., "wimmie262000" <wvh@...> wrote:

--- In BITX20@..., "Allison Parent" <kb1gmx@> wrote:

The Negative tempco as you refer is a side effect of the positive
tempco of the gate threshold (bias point for linear operation).
Could you elaborate on this? I am trying to understand why there is a
negative tempco in the sub threshold region of a power mosfet.

Joop
Simple answer is with increased temperature in silicon there is
increased electron mobility.

This is one parameter you never see on the spec curve as the
devices are designed and exepected to be used at saturation or
cutoff.

Allison

Re: Question for Arv

wimmie262000 <wvh@...>
 

--- In BITX20@..., "Allison Parent" <kb1gmx@...> wrote:

The Negative tempco as you refer is a side effect of the positive
tempco of the gate threshold (bias point for linear operation).
Could you elaborate on this? I am trying to understand why there is a
negative tempco in the sub threshold region of a power mosfet.

Joop

Re: Was: Update on BITX20 Kit Progress - tell what the problem is please?

Jim Strohm <n6otq@...>
 

--- arv evans <arvid.evans@...> wrote:

Hi

There are several mentions in historical posts
regarding instability and
low power output from the BITX20 design. The
original Farhan design
used a separate board (or more open spacing) for the
the RF PA section,
which seems to have been easier to tame than those
(like my personal
first construction) that had the PA section close to
other stages. Some
of the "inherently stable" designs become less than
stable if compressed
into small spaces where unwanted coupling becomes
problematic. The
problem is not in the original design. It is with
compressing the whole
rig onto a very tightly packed PCB layout with no
shielding between
stages.
Thank you, Arv.

In addition, after the first pass of the PCB layout,
we fixed a couple of inevitable early-design glitches,
and also adjusted some component values and placements
to help give a better final product at a lower final
cost.

This kind of tweaking in development is normal in the
real world. The difference from the real word for
BITX is -- with the exception of your humble tech
writer (that would be me), everybody else pretty much
has a regular fulltime job. So all the designers and
the beta builders have to dovetail their BITX
development around their daytime jobs.

My occupation right now is to find a good-paying gig
that supports my next political campaign (2008 season)
and also allows me to pursue work that fosters my
career as a free-lance writer, which includes helping
out on projects like this. So you know -- my "pay" to
help with this project is to help with this project.
Like everybody else except Doug, who has a significant
personal financial risk in this project, I'm working
for free.

So I hope everybody will understand why things have
not been progressing at the normal 40 hours a weel
schedule of industry.

We WILL deliver a product that will delight you. At
least that's been our first and only goal the whole
time.

Jim N6OTQ



____________________________________________________________________________________
Expecting? Get great news right away with email Auto-Check.
Try the Yahoo! Mail Beta.
http://advision.webevents.yahoo.com/mailbeta/newmail_tools.html

Re: Question for Arv

Allison Parent <kb1gmx@...>
 

--- In BITX20@..., arv evans <arvid.evans@...> wrote:

Allison

I wonder if anybody has ever mounted their MOSFETs on a peltier plate
and thus used electronic means to pump heat away from the device? The
weird side of me says that maybe one could put the peltier junctions in
series with power to the linear amp, such that increased RF amp current
would cause increased heat pump action.
Save for a peltier that would move 20-30W of heat would suck up
a lot more than that.

What I've done that helps is use a .125" (3.175mm) copper
plate between the device and the finned aluminum heatsink.
Copper has a better heat transfer (lower thermal R) and
halps a bit. Also the device is sitting in a well. It does
help but the real problem with TO220 style cases is getting
the heat from the die to the mounting tab.

The alternate mindset is: For a dollar or so, if you blow one
up on occasion it's not a great expense.

I don't like that but some feel its valid.


Allison

Arv
_._

Allison Parent wrote:

--- In BITX20@... <mailto:BITX20%40yahoogroups.com>,
"wimmie262000" <wvh@> wrote:

--- In BITX20@... <mailto:BITX20%40yahoogroups.com>,
"wimmie262000" <wvh@> wrote:
I just tried to read up on this phenomena and hoped to find a clear
explanation why this is happening. E.g. see this article:
http://www.onsemi.com/pub/Collateral/AND8199-D.PDF
<http://www.onsemi.com/pub/Collateral/AND8199-D.PDF>
I did not find the most trustworthy source on the details, but this
seems to happen:
* Switching MOSFETS are made of many parallel little MOSFETS.
* At high drive level (fully open), the channel resistance is the
major factor and has a positive tempco.
* At low drive level (near pinch-off), the pinch-off voltage is the
major factor. But this has a negative tempco (why, dunno?).
What we do in our PA's is use a switching MOSFET in "linear region".
So the individual mini mosfets inside will compete around the on/off
range. One could see an individual mosfet then as a potential
"hotspot".

Even this stype of fet does not have a hot spot problem.

The Negative tempco as you refer is a side effect of the positive
tempco of the gate threshold (bias point for linear operation).

From articles on the topic of reliability of some motor drivers and
power supplies I read that is is better to stay away from this
linear
zone.
For switching applications its important as it limits efficientcy.
These FETs are designed for switching applications where large amounts
of power is not dissapated.

For not switching applications such as linear RF amplicfication
it's a contradiction. FYI: there are people using power fets for
Class E AM opertation in the kilorwatt range using the switching
characteristics of these devices to advanatage.

Also not be tempted to select an even bigger MOSFET for a
certain job since the circuit most likely will operate even more
into
the "almost-off zone" and be subject to a bigger negative tempco. If
some linear operation is unavoidable it is better to take a MOSFET
type just enough suited for the job and then cool it as well as
possible.
This last thing is separate from bias temperature adjustment, which
seems necessary anyway.
the liability of using a bigger fet is that for RF serrvice the
capacitances go up significantly and that requires lower operating
impedences to get good operating efficientcy. If the efficientcy
is poor due to bad design the heating problem is made worse.

Bigger fets are the same as smaller fets in general behavour. The
problem is FETs designed for RF service are EXPENSIVE. There are
a number of RF FETs and they are good but devices like the 140W
MRF140 are in the $1US/W price range(I've seen prices between
$90-140 each). Even then a bias scheme that tracks temperature
is deisreable to protect an expensive device.

The appeal of the IRF5xx devices is that they are cheap. For
comparison to RF power fets, IRF510 is around $0.13US/W (roughly
$1.29 each)! Their problems are known and with attention can be
dealt with. The biggest problem I see is the TO-220 style case
has a very high thermal resistance so dissapating high power is
more a problem of transfering the heat. The same die in a more
thermally conductive package would do far better and also help
reduce the effects of bias shift.

If heat transfer was not a problem the IRF510 can easily deliver
far more power. An example of this is the March 1999 QST design
that uses a pair of them push pull with 24V DC at an average power
output of 45W for bands 80M-10M. The author has curves for the
amp in a later issue (April or May 1999) showing as much as 70W
to a load at around 7mhz for this amp. He also comments on the
heat transfer problem. It shows just how far you can push these
devices.

Allison



Re: Was: Update on BITX20 Kit Progress - tell what the problem is please?

Arv Evans
 

Stan

It might be interesting to scroll back to BITX20 message #439. It is
not the only one, but that one is classic.

Arv
_._

--- In BITX20@..., "Stan" <ak0b@...> wrote:

--- In BITX20@..., "KI6DS" <ki6ds@> wrote:

We found that the rig had severe stability
problems as designed
snip<
> informed of our progress. 72, Doug, KI6DS
Doug several hundred of the BITX20 have been built around the world and
this stability problem has not been reported even though vector board,
manhatten style and PCB of several different designs have been laid out
and built by others.

Please explain exactly what the stability problem with the design is to
the group. The feedback amplifiers as noted in Wes's ARRL EMRF book
are unconditional stable. I do not think it is fair to the group or to
Farhan to just say "problems as designed"

Thanks, Stan AK0B

Re: Was: Update on BITX20 Kit Progress - tell what the problem is please?

Arv Evans
 

Hi

There are several mentions in historical posts regarding instability and
low power output from the BITX20 design. The original Farhan design
used a separate board (or more open spacing) for the the RF PA section,
which seems to have been easier to tame than those (like my personal
first construction) that had the PA section close to other stages. Some
of the "inherently stable" designs become less than stable if compressed
into small spaces where unwanted coupling becomes problematic. The
problem is not in the original design. It is with compressing the whole
rig onto a very tightly packed PCB layout with no shielding between
stages. However, we think this is solvable without resorting to a
separate RF PA board. The latest board layout seems to support that
thinking.

In addition to wanting the kit to have a high probably of success for
both neophyte and experienced builders, there is the issue of US
regulations requiring a very strong output filter for suppression of
harmonic energy. Sourcing the kit in the US makes it mandatory that
such filtering be incorporated in the design. This is another reason
for RF output section changes.

In the initial Hendricks kit PCB layout, work was done to find a fix for
the sometimes heard complaint about apparent low received signal level.
We changed the receive section's one-transistor 2nd IF amplifier to a
fixed gain 2-transistor configuration (still preserving the
bi-directional layout). Unfortunately we set the gain too high for the
closely packed board layout. In the second board layout the gain of
this stage has been significantly reduced (builders can change this gain
by changing two resistors if they want to do so), but gain is still more
than was provided by Q-3 in the original schematic on Farhan's web
page. Physical component placement was also changed to minimize any
chance of microphonics from this higher gain. Also, each crystal can is
now grounded in the second board layout to further stabilize the IF section.

Ultimately, this new kit will result in a high quality transceiver. The
original circuit was designed with consideration of parts availability
in India. Since a kit provides all the board level components, some of
the parts availability issues do not apply. Other design
characteristics such as ease of construction and alignment are still
very important, and need to be built into the kit design.

My personal thinking is that Farhan's original design will continue to
evolve and grow over time. There are a number of modifications that
look quite interesting for specific user desires and needs. Some of
these may be incorporated into future re-designs as builder options, or
as external add-on circuits. However, there is the concept of keeping
close enough to Farhan's original design to make the rig suitable for
easy construction from readily available parts. While a parts-provided
kit might allow significant deviation from that concept, it is probably
desirable to attempt maintaining a close similarity between various
kits and the intent of Farhan's original design.

Arv
_._


Jim Strohm wrote:



--- Stan <ak0b@... <mailto:ak0b%40swbell.net>> wrote:

Please explain exactly what the stability problem
with the design is to
the group. The feedback amplifiers as noted in
Wes's ARRL EMRF book
are unconditional stable. I do not think it is fair
to the group or to
Farhan to just say "problems as designed"
Stan,

Doug can provide a more detailed and technically
complete answer -- this is the elevator story.

We found some instability (oscillation) and some
low-power-out issues with the driver and final PA. A
lot of this may have come from the physical layout of
the circuit on the PCB, which is much tighter than the
homebrew construction typically used for BITX.

Rather than try to band-aid the existing circuit, we
decided to go with push-pull driver and PA amps and an
additional pre-driver stage so that the kit would be
unconditionally stable in the final ... and
unconditionally simple for every builder to
successfully assemble.

Because it's no fun to build a radio that doesn't
work, or worse, doesn't work right.

Jim N6OTQ

__________________________________________________________
Finding fabulous fares is fun.
Let Yahoo! FareChase search your favorite travel sites to find flight
and hotel bargains.
http://farechase.yahoo.com/promo-generic-14795097
<http://farechase.yahoo.com/promo-generic-14795097>

Re: Question for Arv

Arv Evans
 

Allison

I wonder if anybody has ever mounted their MOSFETs on a peltier plate
and thus used electronic means to pump heat away from the device? The
weird side of me says that maybe one could put the peltier junctions in
series with power to the linear amp, such that increased RF amp current
would cause increased heat pump action.

Arv
_._

Allison Parent wrote:


--- In BITX20@... <mailto:BITX20%40yahoogroups.com>,
"wimmie262000" <wvh@...> wrote:

--- In BITX20@... <mailto:BITX20%40yahoogroups.com>,
"wimmie262000" <wvh@> wrote:
I just tried to read up on this phenomena and hoped to find a clear
explanation why this is happening. E.g. see this article:
http://www.onsemi.com/pub/Collateral/AND8199-D.PDF
<http://www.onsemi.com/pub/Collateral/AND8199-D.PDF>
I did not find the most trustworthy source on the details, but this
seems to happen:
* Switching MOSFETS are made of many parallel little MOSFETS.
* At high drive level (fully open), the channel resistance is the
major factor and has a positive tempco.
* At low drive level (near pinch-off), the pinch-off voltage is the
major factor. But this has a negative tempco (why, dunno?).
What we do in our PA's is use a switching MOSFET in "linear region".
So the individual mini mosfets inside will compete around the on/off
range. One could see an individual mosfet then as a potential "hotspot".
Even this stype of fet does not have a hot spot problem.

The Negative tempco as you refer is a side effect of the positive
tempco of the gate threshold (bias point for linear operation).

From articles on the topic of reliability of some motor drivers and
power supplies I read that is is better to stay away from this linear
zone.
For switching applications its important as it limits efficientcy.
These FETs are designed for switching applications where large amounts
of power is not dissapated.

For not switching applications such as linear RF amplicfication
it's a contradiction. FYI: there are people using power fets for
Class E AM opertation in the kilorwatt range using the switching
characteristics of these devices to advanatage.

Also not be tempted to select an even bigger MOSFET for a
certain job since the circuit most likely will operate even more into
the "almost-off zone" and be subject to a bigger negative tempco. If
some linear operation is unavoidable it is better to take a MOSFET
type just enough suited for the job and then cool it as well as
possible.
This last thing is separate from bias temperature adjustment, which
seems necessary anyway.
the liability of using a bigger fet is that for RF serrvice the
capacitances go up significantly and that requires lower operating
impedences to get good operating efficientcy. If the efficientcy
is poor due to bad design the heating problem is made worse.

Bigger fets are the same as smaller fets in general behavour. The
problem is FETs designed for RF service are EXPENSIVE. There are
a number of RF FETs and they are good but devices like the 140W
MRF140 are in the $1US/W price range(I've seen prices between
$90-140 each). Even then a bias scheme that tracks temperature
is deisreable to protect an expensive device.

The appeal of the IRF5xx devices is that they are cheap. For
comparison to RF power fets, IRF510 is around $0.13US/W (roughly
$1.29 each)! Their problems are known and with attention can be
dealt with. The biggest problem I see is the TO-220 style case
has a very high thermal resistance so dissapating high power is
more a problem of transfering the heat. The same die in a more
thermally conductive package would do far better and also help
reduce the effects of bias shift.

If heat transfer was not a problem the IRF510 can easily deliver
far more power. An example of this is the March 1999 QST design
that uses a pair of them push pull with 24V DC at an average power
output of 45W for bands 80M-10M. The author has curves for the
amp in a later issue (April or May 1999) showing as much as 70W
to a load at around 7mhz for this amp. He also comments on the
heat transfer problem. It shows just how far you can push these
devices.

Allison

Re: Question for Arv

Arv Evans
 

Hi

There seem to be several possibilities for using MOSFETs that seem to
have not been fully explored:

1) Adding a small Source resistance and bypass, so that increased
idling current would increase the Source voltage and thus effectively
decrease the gate bias. Could some level of DC stability be reached
this way without sacrificing too much power output?

2) I have never seen a Grounded-Gate MOSFET RF amplifier. Is it
possible to use MOSFETs in the equivalent of the old tube type Grounded
Grid Linear Amplifier?

3) Might it be possible to use a MOSFET, or MOSFETs in
Source-Follower mode as an RF amplifier? The inherent feedback should
improve linearity. Drive requirement would be larger, but some drive
energy would be coupled through to the output. If operated from a
negative power supply the Drain lead could then be directly grounded,
along with the heatsink tab on TO-220 devices, making heat dissipation
much easier to deal with.

Arv
_._



wimmie262000 wrote:


--- In BITX20@... <mailto:BITX20%40yahoogroups.com>,
"wimmie262000" <wvh@...> wrote:
I just tried to read up on this phenomena and hoped to find a clear
explanation why this is happening. E.g. see this article:
http://www.onsemi.com/pub/Collateral/AND8199-D.PDF
<http://www.onsemi.com/pub/Collateral/AND8199-D.PDF>
I did not find the most trustworthy source on the details, but this
seems to happen:
* Switching MOSFETS are made of many parallel little MOSFETS.
* At high drive level (fully open), the channel resistance is the
major factor and has a positive tempco.
* At low drive level (near pinch-off), the pinch-off voltage is the
major factor. But this has a negative tempco (why, dunno?).
What we do in our PA's is use a switching MOSFET in "linear region".
So the individual mini mosfets inside will compete around the on/off
range. One could see an individual mosfet then as a potential "hotspot".

From articles on the topic of reliability of some motor drivers and
power supplies I read that is is better to stay away from this linear
zone. Also not be tempted to select an even bigger MOSFET for a
certain job since the circuit most likely will operate even more into
the "almost-off zone" and be subject to a bigger negative tempco. If
some linear operation is unavoidable it is better to take a MOSFET
type just enough suited for the job and then cool it as well as possible.
This last thing is separate from bias temperature adjustment, which
seems necessary anyway.

Re: Was: Update on BITX20 Kit Progress - tell what the problem is please?

Rahul Srivastava
 

Hi!
I feel problems are due to compact all in one board layout. Second use of 2N3904 instead of cheap BC547 for Ver3 layout resulting in higher gain, this however can be reduced by tailoring feedback resistor.
Another reason could be that IRF510 is used without the associated 10ohm gate series resitor that is key to achieveing stability of these linears as analyzed by Wes W7ZOI in one of his writeups. In may units I used IRF530.
I have had assembled about 12+ rigs under my supervision and all were plug and play using the 2 board layout single sided boards.
I am sure Doug/Dan version will better the design, looking forward to detailed analysis.
73
Rahul VU3WJM



ki6ds@... wrote:
Stan, I am not trying to disrespect any of the other work that has been
done. I was trying to say that we had stability problems with our layout.
I am just trying to keep the group informed. Perhaps Arv can speak to
this as Dan is out of the country on vacation. I am not building
prototypes. Dan, Arv and Jim Kortge are.

--- In BITX20@..., "KI6DS" <ki6ds@...> wrote:

We found that the rig had severe stability
problems as designed
snip<
informed of our progress. 72, Doug, KI6DS
Doug several hundred of the BITX20 have been built around the world and
this stability problem has not been reported even though vector board,
manhatten style and PCB of several different designs have been laid out
and built by others.

Please explain exactly what the stability problem with the design is to
the group. The feedback amplifiers as noted in Wes's ARRL EMRF book
are unconditional stable. I do not think it is fair to the group or to
Farhan to just say "problems as designed"

Thanks, Stan AK0B







---------------------------------
All New Yahoo! Mail – Tired of unwanted email come-ons? Let our SpamGuard protect you.

[Non-text portions of this message have been removed]

Re: Was: Update on BITX20 Kit Progress - tell what the problem is please?

Jim Strohm <n6otq@...>
 

--- Stan <ak0b@...> wrote:

Please explain exactly what the stability problem
with the design is to
the group. The feedback amplifiers as noted in
Wes's ARRL EMRF book
are unconditional stable. I do not think it is fair
to the group or to
Farhan to just say "problems as designed"
Stan,

Doug can provide a more detailed and technically
complete answer -- this is the elevator story.

We found some instability (oscillation) and some
low-power-out issues with the driver and final PA. A
lot of this may have come from the physical layout of
the circuit on the PCB, which is much tighter than the
homebrew construction typically used for BITX.

Rather than try to band-aid the existing circuit, we
decided to go with push-pull driver and PA amps and an
additional pre-driver stage so that the kit would be
unconditionally stable in the final ... and
unconditionally simple for every builder to
successfully assemble.

Because it's no fun to build a radio that doesn't
work, or worse, doesn't work right.

Jim N6OTQ



____________________________________________________________________________________
Finding fabulous fares is fun.
Let Yahoo! FareChase search your favorite travel sites to find flight and hotel bargains.
http://farechase.yahoo.com/promo-generic-14795097

Re: Was: Update on BITX20 Kit Progress - tell what the problem is please?

ki6ds@...
 

Stan, I am not trying to disrespect any of the other work that has been
done. I was trying to say that we had stability problems with our layout.
I am just trying to keep the group informed. Perhaps Arv can speak to
this as Dan is out of the country on vacation. I am not building
prototypes. Dan, Arv and Jim Kortge are.

--- In BITX20@..., "KI6DS" <ki6ds@...> wrote:

We found that the rig had severe stability
problems as designed
snip<
> informed of our progress. 72, Doug, KI6DS
Doug several hundred of the BITX20 have been built around the world and
this stability problem has not been reported even though vector board,
manhatten style and PCB of several different designs have been laid out
and built by others.

Please explain exactly what the stability problem with the design is to
the group. The feedback amplifiers as noted in Wes's ARRL EMRF book
are unconditional stable. I do not think it is fair to the group or to
Farhan to just say "problems as designed"

Thanks, Stan AK0B


Was: Update on BITX20 Kit Progress - tell what the problem is please?

myid_63301
 

--- In BITX20@..., "KI6DS" <ki6ds@...> wrote:

We found that the rig had severe stability
problems as designed
snip<
> informed of our progress. 72, Doug, KI6DS
Doug several hundred of the BITX20 have been built around the world and
this stability problem has not been reported even though vector board,
manhatten style and PCB of several different designs have been laid out
and built by others.

Please explain exactly what the stability problem with the design is to
the group. The feedback amplifiers as noted in Wes's ARRL EMRF book
are unconditional stable. I do not think it is fair to the group or to
Farhan to just say "problems as designed"

Thanks, Stan AK0B

Re: Question for Arv

Allison Parent <kb1gmx@...>
 

--- In BITX20@..., "MICHAEL CLARKE" <michaelclarke719@...>
wrote:

In his widely published 1.8 to 10.1MHz 5W Mosfet Amplifier, Drew
Diamond, VK3XU had two IRF510s push-pull, with the bias zener diode
positioned against the heatsinks with a small blob of petroleum jelly
so that it tracked temperature causing bias voltage to go down when
the temperature goes up.
Details in (1) RSGB Handbook, 7th edition, p. 5.22, (2) Drew
Diamond, Amateur Radio 10/88 (3) his book Projects for the Radio
Amateur. There may also be information on the Australian amateur
radio site. Try a Google on VK3XU.
73
Michael Mi5MTC

[Non-text portions of this message have been removed]
You can find info on the HFpacker amp (2W in 35 out for
80m-10M) here:

www.hfprojects.com/projects/hfpacker/HFPackerAmp.pdf

And if you search around you will find a similar design
from WA2EBY.


Allison

Re: Question for Arv

MICHAEL CLARKE <michaelclarke719@...>
 

In his widely published 1.8 to 10.1MHz 5W Mosfet Amplifier, Drew Diamond, VK3XU had two IRF510s push-pull, with the bias zener diode positioned against the heatsinks with a small blob of petroleum jelly so that it tracked temperature causing bias voltage to go down when the temperature goes up.
Details in (1) RSGB Handbook, 7th edition, p. 5.22, (2) Drew Diamond, Amateur Radio 10/88 (3) his book Projects for the Radio Amateur. There may also be information on the Australian amateur radio site. Try a Google on VK3XU.
73
Michael Mi5MTC

Re: Question for Arv

Allison Parent <kb1gmx@...>
 

--- In BITX20@..., "wimmie262000" <wvh@...> wrote:

--- In BITX20@..., "wimmie262000" <wvh@> wrote:
I just tried to read up on this phenomena and hoped to find a clear
explanation why this is happening. E.g. see this article:
http://www.onsemi.com/pub/Collateral/AND8199-D.PDF
I did not find the most trustworthy source on the details, but this
seems to happen:
* Switching MOSFETS are made of many parallel little MOSFETS.
* At high drive level (fully open), the channel resistance is the
major factor and has a positive tempco.
* At low drive level (near pinch-off), the pinch-off voltage is the
major factor. But this has a negative tempco (why, dunno?).
What we do in our PA's is use a switching MOSFET in "linear region".
So the individual mini mosfets inside will compete around the on/off
range. One could see an individual mosfet then as a potential "hotspot".
Even this stype of fet does not have a hot spot problem.

The Negative tempco as you refer is a side effect of the positive
tempco of the gate threshold (bias point for linear operation).


From articles on the topic of reliability of some motor drivers and
power supplies I read that is is better to stay away from this linear
zone.
For switching applications its important as it limits efficientcy.
These FETs are designed for switching applications where large amounts
of power is not dissapated.

For not switching applications such as linear RF amplicfication
it's a contradiction. FYI: there are people using power fets for
Class E AM opertation in the kilorwatt range using the switching
characteristics of these devices to advanatage.

Also not be tempted to select an even bigger MOSFET for a
certain job since the circuit most likely will operate even more into
the "almost-off zone" and be subject to a bigger negative tempco. If
some linear operation is unavoidable it is better to take a MOSFET
type just enough suited for the job and then cool it as well as
possible.
This last thing is separate from bias temperature adjustment, which
seems necessary anyway.
the liability of using a bigger fet is that for RF serrvice the
capacitances go up significantly and that requires lower operating
impedences to get good operating efficientcy. If the efficientcy
is poor due to bad design the heating problem is made worse.

Bigger fets are the same as smaller fets in general behavour. The
problem is FETs designed for RF service are EXPENSIVE. There are
a number of RF FETs and they are good but devices like the 140W
MRF140 are in the $1US/W price range(I've seen prices between
$90-140 each). Even then a bias scheme that tracks temperature
is deisreable to protect an expensive device.

The appeal of the IRF5xx devices is that they are cheap. For
comparison to RF power fets, IRF510 is around $0.13US/W (roughly
$1.29 each)! Their problems are known and with attention can be
dealt with. The biggest problem I see is the TO-220 style case
has a very high thermal resistance so dissapating high power is
more a problem of transfering the heat. The same die in a more
thermally conductive package would do far better and also help
reduce the effects of bias shift.

If heat transfer was not a problem the IRF510 can easily deliver
far more power. An example of this is the March 1999 QST design
that uses a pair of them push pull with 24V DC at an average power
output of 45W for bands 80M-10M. The author has curves for the
amp in a later issue (April or May 1999) showing as much as 70W
to a load at around 7mhz for this amp. He also comments on the
heat transfer problem. It shows just how far you can push these
devices.


Allison

Re: Question for Arv

wimmie262000 <wvh@...>
 

--- In BITX20@..., "wimmie262000" <wvh@...> wrote:
I just tried to read up on this phenomena and hoped to find a clear
explanation why this is happening. E.g. see this article:
http://www.onsemi.com/pub/Collateral/AND8199-D.PDF
I did not find the most trustworthy source on the details, but this
seems to happen:
* Switching MOSFETS are made of many parallel little MOSFETS.
* At high drive level (fully open), the channel resistance is the
major factor and has a positive tempco.
* At low drive level (near pinch-off), the pinch-off voltage is the
major factor. But this has a negative tempco (why, dunno?).
What we do in our PA's is use a switching MOSFET in "linear region".
So the individual mini mosfets inside will compete around the on/off
range. One could see an individual mosfet then as a potential "hotspot".

From articles on the topic of reliability of some motor drivers and
power supplies I read that is is better to stay away from this linear
zone. Also not be tempted to select an even bigger MOSFET for a
certain job since the circuit most likely will operate even more into
the "almost-off zone" and be subject to a bigger negative tempco. If
some linear operation is unavoidable it is better to take a MOSFET
type just enough suited for the job and then cool it as well as possible.
This last thing is separate from bias temperature adjustment, which
seems necessary anyway.