Topics

I7SWX PTT Pop Mute

Giancarlo
 

Hi All,

 

I see some very interesting comments about the LM386 PTT Pops solution I suggested in february.

 My latest e-mail generated some confusion as I reported Pin 1 in place of Pin 7 to KC8WBK. Please let me explain why.

My original schematic suggestion taking to RX +12V via pin 7 was thought as simulating the increase in value of 1.35kohm resistor, between pin 8 and pin1 (gain), to reduce gain of the balance amplifier, due to the increase of current through it.

I generated another change schematic where pin 1 was taken to RX +12V, see drawing V2. Next day I was taken to hospital and stayed away from home, between hospital and rehab center for four months. I was convinced to have uploaded such mod and following my memory I commented this mod when I read KC8WBK complain note.

 I was not too shure the simulation of pins 8-1 resistor change would be enough and I addressed my solution to take pin 1 to +12V. This move will block any signal to reach the push-pull stage as + input from pin 3 would never be amplified. The only “puzzle” of this solution would be if the 1.23kohm resistor can dissipate circa 100mW due to a current expected of 8mA, as the PNP darlington transistors, associated to pin 2 - input connected to ground, have a voltage at the emitter of probably +1.4V. If someone will try and demostrate that is not a valid solution and blows the LM386 I will send him a couples as a replacement.... We are experimening .... hi.

 Best 73

 Gian

I7SWX

KC8WBK
 

Switching lead from pin 7 to pin 1:

PTT pop is gone.
Frequency seems stable on keydown.
No buzz or other background noise during tx.

I will try to get an audio report and will do some on air testing after cleaning up loose ends.

Jerry Gaffke
 

Driving pin 1 to 12v does indeed put about 8ma through the pnp's on the pin 2 side.
More worrisome, nothing much limits the current through the pnp's on the pin 3 side.

Driving pin 1 to something between 0v and 1v does mute the chip quite nicely though.
No significant extra currents.
If we add a cap somewhere to slow down the mute-off, that would have to happen in front of the switch that pulls pin 1 low.

Driving pin 7 to 0v  reduces the voltage gain from 20 to around 3, output is centered on 1.75v.
Driving pin 7 to 12v reduces the voltage gain from 20 to around 10, output is centered on 10.9v.
That second observation of mine in LTSpice does not quite square with the experience of others that this gives nearly full muting.
 
I'm messing around in LTSpice, no need to send any spare LM386's, just a new laptop when I blow this one out.    ;-)
The LTSpice file is here:  https://groups.io/g/BITX20/files/KE7ER/lm386jg.asc
It's a straightforward entry from the internal schematic shown in the LM386 datasheet.
This other Spice model is likely better than mine:  https://forum.allaboutcircuits.com/threads/lm386-ltspice-simulation.121375/

I'm still not confident I understand the LM386's feedback scheme.  The 15k + 1.35k + 150 ohm resistors do effectively form a voltage feedback network into pin 3.
Pin 3 is always 2*Vbe below pin 1, so IN+ at pin 3 effectively sees a feedback voltage that is an inverted version of pin 1.   Also, with pin 2 grounded as it is in the Bitx40,, the left side of that 150 ohm resistor is effectively pegged at an AC ground of 2*Vbe = 1.2vdc.  The datasheet describes the LM386 as an op-amp with internal feedback, and the above is pretty much a standard inverting op-amp configuration with a voltage gain of 15k/(1.35k+150).  However, there is significant current going down into the diff amp from pin 1, which should disturb that model considerably.  And while I still don't quite understand exactly what current would be heading off to the right from the diff-amp to drive the output, that may not matter so long as there is sufficient open loop gain..   

>  The only “puzzle” of this solution would be if the 1.23kohm resistor can dissipate circa 100mW due to a current expected of 8mA
 
I assume you mean the 1.35k resistor.  Anyway, I'd be much more worried about the pnp's on the right hand side of the diff amp if you hit pin 1 with 12v.


Jerry, KE7ER


On Mon, Jul 31, 2017 at 10:57 am, Giancarlo wrote:

My original schematic suggestion taking to RX +12V via pin 7 was thought as simulating the increase in value of 1.35kohm resistor, between pin 8 and pin1 (gain), to reduce gain of the balance amplifier, due to the increase of current through it.

I generated another change schematic where pin 1 was taken to RX +12V, see drawing V2. Next day I was taken to hospital and stayed away from home, between hospital and rehab center for four months. I was convinced to have uploaded such mod and following my memory I commented this mod when I read KC8WBK complain note.

 I was not too shure the simulation of pins 8-1 resistor change would be enough and I addressed my solution to take pin 1 to +12V. This move will block any signal to reach the push-pull stage as + input from pin 3 would never be amplified. The only “puzzle” of this solution would be if the 1.23kohm resistor can dissipate circa 100mW due to a current expected of 8mA, as the PNP darlington transistors, associated to pin 2 - input connected to ground, have a voltage at the emitter of probably +1.4V. If someone will try and demostrate that is not a valid solution and blows the LM386 I will send him a couples as a replacement.... We are experimening .... hi.

 

Giancarlo
 

Hi,

thanks for the update on using pin 1. I am glad and it is like I was expecting.

73

Gian
I7SWX

Giancarlo
 

Hi Jerry,

thanks for your report and data. Yes, 1.35K. Your worry about the transistor driving the push-pull maybe right. KC8WBK has moved from pin 7 to pin 1 and it seems to be working OK. The solution of taking pin 1 to gnd requires an additional NPN transistor, I have attached the V3 schematic.

The LM386 maybe OK as spares for KC8WBK ... for the PC I have an HP with keybod smoked ... hi

73

Gian
I7SWX

Jerry Gaffke
 

Assume an input voltage of 0.1v at pin 3.  Current through the 50k resistor to ground would be 0.1/50000, and assuming a beta of 100 on those two PNP's the maximum current sucked away from pin 1 would be 100*100*0.1/50000 = 0.2ma.  Which is considerably less than the nearly 1ma coming in through the 15k resistor from the output.  So the current down through the PNP's can get significant, but not overpowering.  Pretty clear now why you want an AC coupled source into pin 3.


On Tue, Aug 1, 2017 at 09:56 am, Jerry Gaffke wrote:
The datasheet describes the LM386 as an op-amp with internal feedback, and the above is pretty much a standard inverting op-amp configuration with a voltage gain of 15k/(1.35k+150).  However, there is significant current going down into the diff amp from pin 1, which should disturb that model considerably.

Vince Vielhaber
 

Take a look here:
http://la3za.blogspot.com.au/2003/04/using-pin-7-of-lm386-to-reduce-bci-and.html

Figure 2 shows how to mute the LM386 by grounding pin 7 and feeding pin 6
thru a resistor (10-100 ohms).

Vince.

Driving pin 1 to 12v does indeed put about 8ma through the pnp's on the
pin 2 side.
More worrisome, nothing much limits the current through the pnp's on the
pin 3 side.

Driving pin 1 to something between 0v and 1v does mute the chip quite
nicely though.
No significant extra currents.
If we add a cap somewhere to slow down the mute-off, that would have to
happen in front of the switch that pulls pin 1 low.

Driving pin 7 to 0v  reduces the voltage gain from 20 to around 3, output
is centered on 1.75v.
Driving pin 7 to 12v reduces the voltage gain from 20 to around 10, output
is centered on 10.9v.
That second observation of mine in LTSpice does not quite square with the
experience of others that this gives nearly full muting.
 
I'm messing around in LTSpice, no need to send any spare LM386's, just a
new laptop when I blow this one out.    ;-)
The LTSpice file is here:
 https://groups.io/g/BITX20/files/KE7ER/lm386jg.asc (
https://groups.io/g/BITX20/files/KE7ER/lm386jg.asc )
It's a straightforward entry from the internal schematic shown in the
LM386 datasheet.
This other Spice model is likely better than mine:
 https://forum.allaboutcircuits.com/threads/lm386-ltspice-simulation.121375/
(
https://forum.allaboutcircuits.com/threads/lm386-ltspice-simulation.121375/
)

I'm still not confident I understand the LM386's feedback scheme.  The
15k + 1.35k + 150 ohm resistors do effectively form a voltage feedback
network into pin 3.
Pin 3 is always 2*Vbe below pin 1, so IN+ at pin 3 effectively sees a
feedback voltage that is an inverted version of pin 1.   Also, with pin 2
grounded as it is in the Bitx40,, the left side of that 150 ohm resistor
is effectively pegged at an AC ground of 2*Vbe = 1.2vdc.  The datasheet
describes the LM386 as an op-amp with internal feedback, and the above is
pretty much a standard inverting op-amp configuration with a voltage gain
of 15k/(1.35k+150).  However, there is significant current going down
into the diff amp from pin 1, which should disturb that model
considerably.  And while I still don't quite understand exactly what
current would be heading off to the right from the diff-amp to drive the
output, that may not matter so long as there is sufficient open loop
gain..   

 The only “puzzle� of this solution would be if the 1.23kohm
resistor can dissipate circa 100mW due to a current expected of 8mA
 
I assume you mean the 1.35k resistor.  Anyway, I'd be much more worried
about the pnp's on the right hand side of the diff amp if you hit pin 1
with 12v.

Jerry, KE7ER

On Mon, Jul 31, 2017 at 10:57 am, Giancarlo wrote:




My original schematic suggestion taking to RX +12V via pin 7 was thought
as simulating the increase in value of 1.35kohm resistor, between pin 8
and pin1 (gain), to reduce gain of the balance amplifier, due to the
increase of current through it.



I generated another change schematic where pin 1 was taken to RX +12V,
see
drawing V2. Next day I was taken to hospital and stayed away from home,
between hospital and rehab center for four months. I was convinced to
have
uploaded such mod and following my memory I commented this mod when I
read
KC8WBK complain note.



 I was not too shure the simulation of pins 8-1 resistor change would
be
enough and I addressed my solution to take pin 1 to +12V. This move will
block any signal to reach the push-pull stage as + input from pin 3
would
never be amplified. The only “puzzle� of this solution would be if
the
1.23kohm resistor can dissipate circa 100mW due to a current expected of
8mA, as the PNP darlington transistors, associated to pin 2 - input
connected to ground, have a voltage at the emitter of probably +1.4V. If
someone will try and demostrate that is not a valid solution and blows
the
LM386 I will send him a couples as a replacement.... We are experimening
.... hi.



 

Andrew Kasurak
 

why do we not just use a transistor to shunt pin 5 to ground when TX is high? Since the 386 will be disabled and only running on capacitors there shouldn't be too much energy to move.

Jerry Gaffke
 

That's pretty much what Gian started out with.
Gian was pulling pin 7 high, the article says hi or lo works.

As I say, lots of others seem to be finding that grounding pin 7 (or tying it to VCC) gives a full mute.  My simulation says grounding pin 1 is more effective.
But I still don't fully understand everything going on here.

It could be that we could have a cap to ground on pin 1 to slow the unmute, it would also increase the gain much like a cap between pins 1 and 8.  But probably needs a series resistor to keep the gain from getting out of hand.
 
Jerry


On Tue, Aug 1, 2017 at 10:52 am, Vince Vielhaber wrote:
Figure 2 shows how to mute the LM386 by grounding pin 7 and feeding pin 6
thru a resistor (10-100 ohms).

Jerry Gaffke
 

Do try it if you feel so inclined.  Should work.  Though I'm inclined to mess with either pin 1 or pin 7, as that avoids the cap discharge you mention, and is useful in other situations beyond the Bitx40.


On Tue, Aug 1, 2017 at 11:08 am, Andrew Kasurak wrote:
why do we not just use a transistor to shunt pin 5 to ground when TX is high? Since the 386 will be disabled and only running on capacitors there shouldn't be too much energy to move.

KC8WBK
 

Regarding version 2:

I am using amunters v14.  The frequency changes happen(ed) as the radio moves from rx to tx, and back from tx to rx.  During TX there is no change, it is only as you keydown and keyup.  I was not able to download v20 this afternoon, perhaps internet connection problems.

Disconnecting the PTT mod C150 to the black wire of the Raduino provided relief from frequency changes. 
When I reconnected this wire, the frequency remained stable.  ???
I then added filter caps, 0.1uF across the Raduino power regulator, both legs to ground.  I also added a cap across violet to yellow on the tuning pot, which I neglected when I replaced the original with a 10 turn pot.

So far, the pop mute works well and the frequency is now stable.  I will need some good band conditions and an evening of making noise on the bands to say that the board is stable and suitable for recommendation to all.  But at this point, it seems quite positive.

I had an idea that did not work: I thought I would make a simple AGC that would take input from LM386 pin 3, run it through a diode and variable pot to the base of a 2N4403, with the collector to positive 12V and emitter through a 1.2k resistor to LM386 pin 1.  It shut the audio down and the sketch was corrupted at this time (related?).  It might be possible to shave the audio peaks or work as an actual AGC.  I don't really know what I am doing so I would have to learn a lot to design this, which would be fun. One thing that I discovered is that it does not take 12V to mute the LM386 pin 1.  I put a measured 4V on pin 1 and audio was shut down. Pin 1 seems to run about 1.5V under normal operation.

I may try this mod but I decided to work on the VK3YE AGC first, because I have it ready to be installed and I have great hope that it may actually work this time.

KC8WBK
 

I have npn transistors on order and will try version 3 below when I get them from Thailand.

Is the problem with taking pin 1 to 12V too much current?

Jerry Gaffke
 

Too much current.
Goes down through the two transistors on the right side of the diff amp.
From emitter to base on the upper one, then from emitter to base at pin 3 and (mostly) emitter to collector (grounded) on the lower transistor.
Possibly several amps, at least for a few milliseconds.
If you have a beefy enough power supply.

Jerry


On Thu, Aug 3, 2017 at 12:21 pm, KC8WBK wrote:
I have npn transistors on order and will try version 3 below when I get them from Thailand.

Is the problem with taking pin 1 to 12V too much current?

Giancarlo
 

,

probably the best is take pin 1 to gnd.
If you send me your address I can mail you some NPN PNP  transistors SMD or standard?

73

I7SWX
Gian

KC8WBK
 

That is a very kind offer but I am not in a hurry, so I can wait until the package arrives from Futurelec.

Giancarlo
 

OK, no problems.

73

Gian
I7SWX

KC8WBK
 

I just want to confirm that version 3 works.  It functions the same as previous versions, but draws LM386 pin 1 down to ground rather than up to 12V, due to concerns about current in the LM386.  I substituted a 2N3904 for the 2N2222, otherwise identical with the schematic here: https://groups.io/g/BITX20/attachment/30784/0/BITX_40%20PTT%20CLICKS%20V3.png

I had a question about the board and how it was built.  It is called manhattan-style, with raised pads punched from copper clad board with a hole punch, glued to the copper clad board used as the ground plane.
 
Here is a pdf of Manhattan Building Techniques, K7QO
http://www.unixnut.net/files/manart.pdf 

Giancarlo
 

Hi, I have loaded a file with pictures of my assembled board. Also, I am adding the file to this reply.

You can see my "legs up" way.

73
Gian
I7SWX

KN4AAG Kevin Estep
 

I'm planning to add this over the weekend. Thanks to everyone for all of the incredible ideas and mods, both hardware and software.

(Hopefully back on topic now...)

73, Kevin, KN4AAG

Jack Purdum <econjack@...>
 

Is there a schematic for this two transistor version? I only see the one transistor version.

Jack, W8TEE



From: Giancarlo via Groups.Io <i7swx@...>
To: BITX20@groups.io
Sent: Friday, August 11, 2017 12:58 PM
Subject: Re: [BITX20] I7SWX PTT Pop Mute

Hi, I have loaded a file with pictures of my assembled board. Also, I am adding the file to this reply.

You can see my "legs up" way.

73
Gian
I7SWX



Virus-free. www.avast.com