Topics

Protection of Q90

Gordon Gibby
 

I carefully soldered a diode reverse direction across the base emitter junction of Q 90.  Used nearby passive components as places to make the connection.

Haven't heard anything unusual, made multiple winlink connections afterwards with 5 W output.  Still having RFI problems above  that power level. (The unit has virtually no shielding The way I built it.)

Workin great!!


image1.JPG

Sent from my iPhone

Al Duncan VE3RRD
 

Both Q13 in the Bitx40 and Q90 in the uBitx are the first stage in the transmit chain. Both have the "issue" of remaining connected to the antenna port when the transceiver is in receive mode. In both radios, if protection diodes were to be used, they should be installed in the receive part of the circuit (relay K1 pin 12 to ground). That way they are only in the circuit during receive and are shorted to ground during transmit. Placing a diode in the transmit circuit could generate undesirable spurs or other products that can be amplified by the following transmit stages.
Ashhar has mentioned that the uBitx is quite "bullet proof" and shouldn't need diode protection against strong nearby transmitters.

AL VE3RRD

Jerry Gaffke
 

Here's Farhan's post from  https://groups.io/g/BITX20/message/35656
>  the front end diodes in the mixer will clamp any signal higher than half a volt.
>  this should take care of the very powerful signals.
>  adding diodes at the open front end can lead to a very large number of spurs in ubitx.
>  unfortunately, unlike the bitx40, ubitx does not have the luxury of a narrow, triple tunec circuit.

Since the uBitx does not have a bandpass filter before that first mixer, 
adding a non-linear device like those diodes up front will create much more cruft
in the uBitx receiver than it does in the Bitx40.
We normally think of a silicon diode as not conducting up to a knee at around 0.6v,
but in reality the current is an exponential function of voltage starting at very low voltages.
So even some moderately strong AM broadcast stations could have harmonics
and unpredictable mixing products from those added diodes that would interfere with
the reception of weak ham band signals.

Gordon might be right about the potential failure mode in Q90.
But I think Farhan is right that strong incoming signals up to 30mhz will be clipped
by the BAT54s diodes D1 and D2 before any such signal becomes a problem at Q90.
Note that T2 steps up the voltage to D1 and D2 by a factor of 2 due to the winding ratio,
and whether its a positive or negative RF peak there will be two schottky diode drops
limiting that peak voltage to around 0.3+0.3=0.6v at D1,D2, or 0.3v as seen by Q90.

The low pass filter at L1,2,3,4 has no effect on signals below 30mhz, but blocks VHF.
So possible that a strong local VHF transmission could blow Q90.
If that's a concern, an easy fix is to have the uBitx receiver make use of the transmit LPF's. 

All of this is worth a closer examination, perhaps some experimenting.

Jerry, KE7ER

Gordon Gibby
 

"Since the uBitx does not have a bandpass filter before that first mixer, 
adding a non-linear device like those diodes up front will create much more cruft
in the uBitx receiver than it does in the Bitx40.
We normally think of a silicon diode as not conducting up to a knee at around 0.6v,
but in reality the current is an exponential function of voltage starting at very low voltages.
So even some moderately strong AM broadcast stations could have harmonics
and unpredictable mixing products from those added diodes that would interfere with
the reception of weak ham band signals.

Gordon might be right about the potential failure mode in Q90.
But I think Farhan is right that strong incoming signals up to 30mhz will be clipped
by the BAT54s diodes D1 and D2 before any such signal becomes a problem at Q90.
Note that T2 steps up the voltage to D1 and D2 by a factor of 2 due to the winding ratio,
and whether its a positive or negative RF peak there will be two schottky diode drops
limiting that peak voltage to around 0.3+0.3=0.6v at D1,D2, or 0.3v as seen by Q90.

The low pass filter at L1,2,3,4 has no effect on signals below 30mhz, but blocks VHF.
So possible that a strong local VHF transmission could blow Q90."


If that's a concern, an easy fix is to have the uBitx receiver make use of the transmit LPF's. 



Yes, so very well aware of Ashhar's commentary and your discussion of VHF signals is quite appropriate (as I have a 2 meter packet repeater in the same room with the antenna only 10 feet higher....

The partial solution that I've chosen (a reverse diode across the base-emitter of Q90) I would argue
a) makes minimal worsening of any undesired mixing impacts
b) may provide additional protection to Q90 from signals that are not clipped by the shottky's


Mind you, I don't have any impulse response simulation software to apply here.   However (a) there already IS a diode effectively connected to the input circuitry, that being the base-emitter junction of Q90 itself. C80 effectively connects it across the input signal line in Receive.   If your small-signal mixing argument is significant, then we would already have those mixing products, right?  So our issue is to determine whether we will make them signficantly worse, right?

Is there any risk at all, given your argument for the protection from the Shottky's, or is (b) moot?  :  The risk to Q90 is probably not the forward voltage, but instead the reverse voltage which might damage the base-emitter through excessive dissipation as it breaks down (zener like) if the incoming voltage is significant, fast enough, and not well coupled enough to the Shottky's so as to be reduced to safe levels on the input antenna line....   we both agree that indeed, VHF is one such possibility (which as you point out could be dealt with using a LPF at the outside of the entire transceiver.   Due to delays across the low pass filter, there may be others if there is a repetitive pulsed waveform.   I'm not bright enough to be able to prove that one way or the other.    So I believe (b) is not moot until someone can demonstrate that there are no waveforms remaining that are possibly damaging. 

However, placing a reverse diode across the BE junction of Q90 probably protects it additionally against any incoming voltages that ARENT sufficiently clipped by the shottky's.....  thus asserting (b) above,   

and give that there is already ONE diode there, would the mixing products get any worse than 3db stronger?  If adding a second diode to the system cannot be shown to make the mixing products any more than 3dB worse than they ALREADY MIGHT BE (due to the existence of one diode already in the system) then I assert that (a) is proven.


Make any sense?
Gordon






From: BITX20@groups.io <BITX20@groups.io> on behalf of Jerry Gaffke via Groups.Io <jgaffke@...>
Sent: Tuesday, December 26, 2017 11:59 AM
To: BITX20@groups.io
Subject: Re: [BITX20] Protection of Q90
 
Here's Farhan's post from  https://groups.io/g/BITX20/message/35656
>  the front end diodes in the mixer will clamp any signal higher than half a volt.
>  this should take care of the very powerful signals.
>  adding diodes at the open front end can lead to a very large number of spurs in ubitx.
>  unfortunately, unlike the bitx40, ubitx does not have the luxury of a narrow, triple tunec circuit.

Since the uBitx does not have a bandpass filter before that first mixer, 
adding a non-linear device like those diodes up front will create much more cruft
in the uBitx receiver than it does in the Bitx40.
We normally think of a silicon diode as not conducting up to a knee at around 0.6v,
but in reality the current is an exponential function of voltage starting at very low voltages.
So even some moderately strong AM broadcast stations could have harmonics
and unpredictable mixing products from those added diodes that would interfere with
the reception of weak ham band signals.

Gordon might be right about the potential failure mode in Q90.
But I think Farhan is right that strong incoming signals up to 30mhz will be clipped
by the BAT54s diodes D1 and D2 before any such signal becomes a problem at Q90.
Note that T2 steps up the voltage to D1 and D2 by a factor of 2 due to the winding ratio,
and whether its a positive or negative RF peak there will be two schottky diode drops
limiting that peak voltage to around 0.3+0.3=0.6v at D1,D2, or 0.3v as seen by Q90.

The low pass filter at L1,2,3,4 has no effect on signals below 30mhz, but blocks VHF.
So possible that a strong local VHF transmission could blow Q90.
If that's a concern, an easy fix is to have the uBitx receiver make use of the transmit LPF's. 

All of this is worth a closer examination, perhaps some experimenting.

Jerry, KE7ER

Gordon Gibby
 

In simple terms, I'm arguing that adding a second diode to a front end that already has ONE diode

a) doesn't make matters any greatly worse for mixing products and

b) MAY add protection against losing your rig in an unfortunate event at a field day, near an unexpected transmitting antenna, or due to unusual waveforms that aren't steady state.   ​The added protection may be minimal, but it is probably finite, and the added DAMAGE isn't likely to be any greater than 3db. 


So it is minimally costly protection and possible significant protection against an unlikely but possibly catastrophic event.


Sort of the way insurance is:  it damages your financial status slightly, but it is nice to have when  you discover something unexpected has happened.


The LPF at the front end would be an alternative idea worth pursuing for those not wishing to add a 2nd diode....a very good suggestion that I hadn't thought of.



gordon




From: BITX20@groups.io <BITX20@groups.io> on behalf of Gordon Gibby <ggibby@...>
Sent: Tuesday, December 26, 2017 12:41 PM
To: BITX20@groups.io
Subject: Re: [BITX20] Protection of Q90
 

"Since the uBitx does not have a bandpass filter before that first mixer, 
adding a non-linear device like those diodes up front will create much more cruft
in the uBitx receiver than it does in the Bitx40.
We normally think of a silicon diode as not conducting up to a knee at around 0.6v,
but in reality the current is an exponential function of voltage starting at very low voltages.
So even some moderately strong AM broadcast stations could have harmonics
and unpredictable mixing products from those added diodes that would interfere with
the reception of weak ham band signals.

Gordon might be right about the potential failure mode in Q90.
But I think Farhan is right that strong incoming signals up to 30mhz will be clipped
by the BAT54s diodes D1 and D2 before any such signal becomes a problem at Q90.
Note that T2 steps up the voltage to D1 and D2 by a factor of 2 due to the winding ratio,
and whether its a positive or negative RF peak there will be two schottky diode drops
limiting that peak voltage to around 0.3+0.3=0.6v at D1,D2, or 0.3v as seen by Q90.

The low pass filter at L1,2,3,4 has no effect on signals below 30mhz, but blocks VHF.
So possible that a strong local VHF transmission could blow Q90."


If that's a concern, an easy fix is to have the uBitx receiver make use of the transmit LPF's. 



Yes, so very well aware of Ashhar's commentary and your discussion of VHF signals is quite appropriate (as I have a 2 meter packet repeater in the same room with the antenna only 10 feet higher....

The partial solution that I've chosen (a reverse diode across the base-emitter of Q90) I would argue
a) makes minimal worsening of any undesired mixing impacts
b) may provide additional protection to Q90 from signals that are not clipped by the shottky's


Mind you, I don't have any impulse response simulation software to apply here.   However (a) there already IS a diode effectively connected to the input circuitry, that being the base-emitter junction of Q90 itself. C80 effectively connects it across the input signal line in Receive.   If your small-signal mixing argument is significant, then we would already have those mixing products, right?  So our issue is to determine whether we will make them signficantly worse, right?

Is there any risk at all, given your argument for the protection from the Shottky's, or is (b) moot?  :  The risk to Q90 is probably not the forward voltage, but instead the reverse voltage which might damage the base-emitter through excessive dissipation as it breaks down (zener like) if the incoming voltage is significant, fast enough, and not well coupled enough to the Shottky's so as to be reduced to safe levels on the input antenna line....   we both agree that indeed, VHF is one such possibility (which as you point out could be dealt with using a LPF at the outside of the entire transceiver.   Due to delays across the low pass filter, there may be others if there is a repetitive pulsed waveform.   I'm not bright enough to be able to prove that one way or the other.    So I believe (b) is not moot until someone can demonstrate that there are no waveforms remaining that are possibly damaging. 

However, placing a reverse diode across the BE junction of Q90 probably protects it additionally against any incoming voltages that ARENT sufficiently clipped by the shottky's.....  thus asserting (b) above,   

and give that there is already ONE diode there, would the mixing products get any worse than 3db stronger?  If adding a second diode to the system cannot be shown to make the mixing products any more than 3dB worse than they ALREADY MIGHT BE (due to the existence of one diode already in the system) then I assert that (a) is proven.


Make any sense?
Gordon






From: BITX20@groups.io <BITX20@groups.io> on behalf of Jerry Gaffke via Groups.Io <jgaffke@...>
Sent: Tuesday, December 26, 2017 11:59 AM
To: BITX20@groups.io
Subject: Re: [BITX20] Protection of Q90
 
Here's Farhan's post from  https://groups.io/g/BITX20/message/35656
>  the front end diodes in the mixer will clamp any signal higher than half a volt.
>  this should take care of the very powerful signals.
>  adding diodes at the open front end can lead to a very large number of spurs in ubitx.
>  unfortunately, unlike the bitx40, ubitx does not have the luxury of a narrow, triple tunec circuit.

Since the uBitx does not have a bandpass filter before that first mixer, 
adding a non-linear device like those diodes up front will create much more cruft
in the uBitx receiver than it does in the Bitx40.
We normally think of a silicon diode as not conducting up to a knee at around 0.6v,
but in reality the current is an exponential function of voltage starting at very low voltages.
So even some moderately strong AM broadcast stations could have harmonics
and unpredictable mixing products from those added diodes that would interfere with
the reception of weak ham band signals.

Gordon might be right about the potential failure mode in Q90.
But I think Farhan is right that strong incoming signals up to 30mhz will be clipped
by the BAT54s diodes D1 and D2 before any such signal becomes a problem at Q90.
Note that T2 steps up the voltage to D1 and D2 by a factor of 2 due to the winding ratio,
and whether its a positive or negative RF peak there will be two schottky diode drops
limiting that peak voltage to around 0.3+0.3=0.6v at D1,D2, or 0.3v as seen by Q90.

The low pass filter at L1,2,3,4 has no effect on signals below 30mhz, but blocks VHF.
So possible that a strong local VHF transmission could blow Q90.
If that's a concern, an easy fix is to have the uBitx receiver make use of the transmit LPF's. 

All of this is worth a closer examination, perhaps some experimenting.

Jerry, KE7ER

Jerry Gaffke
 

I think you are mostly right.

Q90 during RX has no collector voltage, so it is just a Vbe diode.
Adding back-to-back diodes to ground as was done on the Bitx40 likely does increase receiver hash on the uBitx,
it seems that Farhan may have already confirmed this.
But a reverse diode across the Q90 Vbe junction is at least somewhat tempered by the 
the R/C network from the Q90 emitter to ground.


>  If your small-signal mixing argument is significant, then we would already have those mixing products, right? 

The mixer is driven hard by a square wave from the Si5351,
two of the diodes at D1,D2 are always conducting and should look like resistors
to the incoming RF.  At least for a first order analysis.

Worth some experimentation
Not stuff I have a lot of experience with.

Jerry,  KE7ER

 


On Tue, Dec 26, 2017 at 09:41 am, Gordon Gibby wrote:
The partial solution that I've chosen (a reverse diode across the base-emitter of Q90) I would argue
a) makes minimal worsening of any undesired mixing impacts
b) may provide additional protection to Q90 from signals that are not clipped by the shottky's
 

Gordon Gibby
 

I don't have much experience AT ALL.    I'm a vacuum tubes fella!!!!!


But I try to add at least a whiff of EMP protection to every rig that I use since theoretically our ARES group is supposed to be ready for "emergencies"


QST suggested gas discharge tubes across the antenna lead -- which only work provided the fidelity of the transmission line system is NOT good into the Gigahertz (exceeding the turnon timing of the gas discharge tube) ....but then most ham ssystems are lossy there and slow down the incident waveform enough to give the gas discharge aa chance....then the diode protection and shottky protection have the possibillities of protecting the rig.....


Your idea of getting one of the ALREADY THERE low pass filters to work is a really good one.....I hadn't recognized what you were saying at first.  Perhaps that can be arranged?    One can always put an external LPF to shut off the VHF (and that possibly worth my doing!)  


I'm really amazed at how "finished" the ubitx is, right out of the box.   There is the POP when going into transmit and I havn't made your improvement there yet (soon!)  but it was right on frequency, enough for me to make WINLINK connections, "as is."    


From: BITX20@groups.io <BITX20@groups.io> on behalf of Jerry Gaffke via Groups.Io <jgaffke@...>
Sent: Tuesday, December 26, 2017 1:23 PM
To: BITX20@groups.io
Subject: Re: [BITX20] Protection of Q90
 
I think you are mostly right.

Q90 during RX has no collector voltage, so it is just a Vbe diode.
Adding back-to-back diodes to ground as was done on the Bitx40 likely does increase receiver hash on the uBitx,
it seems that Farhan may have already confirmed this.
But a reverse diode across the Q90 Vbe junction is at least somewhat tempered by the 
the R/C network from the Q90 emitter to ground.


>  If your small-signal mixing argument is significant, then we would already have those mixing products, right? 

The mixer is driven hard by a square wave from the Si5351,
two of the diodes at D1,D2 are always conducting and should look like resistors
to the incoming RF.  At least for a first order analysis.

Worth some experimentation
Not stuff I have a lot of experience with.

Jerry,  KE7ER

 

On Tue, Dec 26, 2017 at 09:41 am, Gordon Gibby wrote:
The partial solution that I've chosen (a reverse diode across the base-emitter of Q90) I would argue
a) makes minimal worsening of any undesired mixing impacts
b) may provide additional protection to Q90 from signals that are not clipped by the shottky's
 

Jerry Gaffke
 

The GDT's fire up around 100v, so any diode protection may have some work to do.
Series resistance may help, Raj's original back-to-back diode suggestion included a small incandescent lamp.

The idea was originally from Raj:  https://groups.io/g/BITX20/message/19105
His incandescent bulb increases it's resistance drastically when it heats up, so is worth considering in addition to the diodes.
https://groups.io/g/BITX20/message/21901

Jerry, KE7ER


On Tue, Dec 26, 2017 at 11:02 am, Gordon Gibby wrote:

QST suggested gas discharge tubes across the antenna lead -- which only work provided the fidelity of the transmission line system is NOT good into the Gigahertz (exceeding the turnon timing of the gas discharge tube) ....but then most ham ssystems are lossy there and slow down the incident waveform enough to give the gas discharge aa chance....then the diode protection and shottky protection have the possibillities of protecting the rig.....

 

Gordon Gibby
 

Yes, the diodes are important.   


However, you can actually get gas discharge tubes that will spark over at 60 volts (nominally) and up to 10kAmperes


https://www.digikey.com/product-detail/en/bourns-inc/2020-15T-C2LF/2020-15T-C2LF-ND/2595617


The E1 wave of a EMP is over in maybe a few nanoseconds so the impact of the series filament would be its impedance to the higher frequency components.   


EMP has a spectrum from DC to about 100 MHz, and then declining to about 1 GHZ.   At least that is what the MIL SPECs say.....


So it is just this incredibly strong radio frequency interference.


Here' I'm speaking about the E1 and E2 waves.   The E3 event is of risk to the power grid, not so much to radios except via losing their power.  


If all we had to do were protect the receivers, it would  be easy.   protecting the TRANSMITTERS is more difficult becuase you just can't put a couple diodes cross the transmitter output!!!!!    


There vacuum tubes have an advantage, and indeed they sailed through the goverment testing published in QST.   


Howver, with the proper filtering, there is still a chance for units like mosfets with 100V ratings used at only 12-24 volts.   If the incoming wave is sufficiently filtered down to a slower rise time, a modest protection with Gas Discharge may suffice.    


It is sorta difficult to test, however......




From: BITX20@groups.io <BITX20@groups.io> on behalf of Jerry Gaffke via Groups.Io <jgaffke@...>
Sent: Tuesday, December 26, 2017 2:13 PM
To: BITX20@groups.io
Subject: Re: [BITX20] Protection of Q90
 
The GDT's fire up around 100v, so any diode protection may have some work to do.
Series resistance may help, Raj's original back-to-back diode suggestion included a small incandescent lamp.

The idea was originally from Raj:  https://groups.io/g/BITX20/message/19105
His incandescent bulb increases it's resistance drastically when it heats up, so is worth considering in addition to the diodes.
https://groups.io/g/BITX20/message/21901

Jerry, KE7ER


On Tue, Dec 26, 2017 at 11:02 am, Gordon Gibby wrote:

QST suggested gas discharge tubes across the antenna lead -- which only work provided the fidelity of the transmission line system is NOT good into the Gigahertz (exceeding the turnon timing of the gas discharge tube) ....but then most ham ssystems are lossy there and slow down the incident waveform enough to give the gas discharge aa chance....then the diode protection and shottky protection have the possibillities of protecting the rig.....

 

Jerry Gaffke
 

Yup, difficult to test.
If I were worried about this sort of thing, I'd dig a fallout shelter and put a couple uBitx's in there all wrapped up in tinfoil.
Unlike NORAD, I don't have to remain on the air during the event.

What about something like a Carrington Event?   https://en.wikipedia.org/wiki/Solar_storm_of_1859
Would protective measures against that be similar?

We get a lot of lightning storms here.
I try to unplug the gear, but then that may not always happen soon enough.
Would be tough for the rig to survive a direct strike to the antenna.
But when a storm's anywhere in the area there are significant electrostatic fields
that should be possible to deal with.
I assume the GDT and some diode scheme would be part of any modern attempt.
A big knife switch nailed to the wall would look kind of cool.

If we must have extra diodes and if they are found to create hash, then perhaps have them at the
antenna port of some external antenna tuner.

Alternately, put diodes in front of some bandpass filter such as this, again outside the uBitx:
http://www.kitsandparts.com/PA3AKE_filter.php
https://martein.home.xs4all.nl/pa3ake/hmode/bpf_all.html

Jerry


On Tue, Dec 26, 2017 at 12:37 pm, Gordon Gibby wrote:

The E1 wave of a EMP is over in maybe a few nanoseconds so the impact of the series filament would be its impedance to the higher frequency components.   

 

Gordon Gibby
 

1.  Using faraday shields to protect "extra" gear isn't really a solution --- because you would realize the potential for an additional attack, and you would never be willing to open up your protected gear.  A principal is:  the gear must be usable in spite of EMP.   QST published articles demonstrating this was achieveable.  


2.  Carrington Events are similar to an E3 from and EMPand basically only involve power line disturbances.   The extensive measures to protect against the nanosecond pulse are unnecessary.   A backup power source (solar, generator etc) is the appropriate solution (as well as an ability to grow your own food).


3.  While nothing stops a direct hit of lightning that WELDS GEAR.....local strikes can be ameliorated by gas discharge, etc.   Current expensive lightning protection gear has $3 gas discharge tubes inside.....


4.  I don't have a good way to measure any change.   I didn't NOTICE any either --- as was pointed out, generally the Shottky's will fire first if there is a steady tone.   However, I have FOUR computer controlled transceivers at my house due to extensive WINLINK server duties...so very important to have some protection for peace of mind.   


Ive found this book to be helpful:  https://www.amazon.com/EMP-Hardened-Radio-Communications-William-Prepperdoc/dp/154077760X



Cheers,


gordon 


From: BITX20@groups.io <BITX20@groups.io> on behalf of Jerry Gaffke via Groups.Io <jgaffke@...>
Sent: Tuesday, December 26, 2017 4:20 PM
To: BITX20@groups.io
Subject: Re: [BITX20] Protection of Q90
 
Yup, difficult to test.
If I were worried about this sort of thing, I'd dig a fallout shelter and put a couple uBitx's in there all wrapped up in tinfoil.
Unlike NORAD, I don't have to remain on the air during the event.

What about something like a Carrington Event?   https://en.wikipedia.org/wiki/Solar_storm_of_1859
Would protective measures against that be similar?

We get a lot of lightning storms here.
I try to unplug the gear, but then that may not always happen soon enough.
Would be tough for the rig to survive a direct strike to the antenna.
But when a storm's anywhere in the area there are significant electrostatic fields
that should be possible to deal with.
I assume the GDT and some diode scheme would be part of any modern attempt.
A big knife switch nailed to the wall would look kind of cool.

If we must have extra diodes and if they are found to create hash, then perhaps have them at the
antenna port of some external antenna tuner.

Alternately, put diodes in front of some bandpass filter such as this, again outside the uBitx:
http://www.kitsandparts.com/PA3AKE_filter.php
https://martein.home.xs4all.nl/pa3ake/hmode/bpf_all.html

Jerry

On Tue, Dec 26, 2017 at 12:37 pm, Gordon Gibby wrote:

The E1 wave of a EMP is over in maybe a few nanoseconds so the impact of the series filament would be its impedance to the higher frequency components.