[BUILD] fripholm's TG1 Zener Limiter boards - support thread

[thelivingroom]

Hey folks, I've spent hours experimenting with adjusting these circuits but I'm still struggling with the calibration on this. The compression curves refuse to match between channels.

Is anyone able to explain exactly what each of the trimmers does? I've not added a trimmer for FT2A as the general consensus is that it's not necessary, but I'm starting to wonder if it will help.

I've attached graphs of my attempts at matching the response curves between channels using a 1kHz sine wave (note colours are not consistent between graphs). So far I've only tried making adjustments in compression mode:

- Graph 1 shows the response when both channels are set with every trimmer at the same resistance and unlinked.

- Graph 2 shows the effects of adjusting FT2B 1 turn and 1.5 turns clockwise, and adjusting the input of R to compensate for gain changes (-1 means one step back from full).

- Graph 3 shows the effects of adjusting FT1 and R input to compensate.

@thelivingroom your build looks great. Hoping I can get there soon... Any chance you could explain your calibration process in more detail?

First I did Fine Tuning Step 1, then didn’t touch that again. I looked at it both on the scope and recording the drum track back into Pro Tools to double check. After that I tried doing Step 2 per the original instructions of adjusting R22 and R56 (mine are older V2.1 boards with trimmers I installed in those resistor spots) with the 1k tone, but the compression was all out of whack, both on the meters and to my ears, after the THD sounded right. With the power off those trimmers can be measured in place, so it was easy to reset and try again a few different ways, but I never could get it to where I was totally satisfied doing it that way. I also noticed when listening to the compression with the stock values of 12k (R22) and 1k (R56) before touching the THD tuning that Ch1 would drop out when the signal got low enough. It sounded like a gate cutting off early. I then decided since compression was more important to me than THD precision that I’d just worry about that and let the THD fall however it ended up. I remembered a couple of mentions in here about lowering R56 to 200R and starting there, so that’s what I eventually did to begin the process again. I also decided to loop a nice sounding, fairly basic and slow, uncompressed mono drum overhead track that had a good decaying crash at the end instead of 1k tone, so I could hear if it was still cutting out. I multed that to both channels so I could also listen for symmetry (panned L and R on my console). Because I could measure the trimmers in place (power off!), I could raise and lower them the same amounts on both channels as I went along. I set R56 to 200R, listened, and compression was good but had no THD. I then raised both channels (R56 only) 100R at a time until the THD became nice and audible but the compression was still clean and mostly symmetrical from fastest to slowest attack and release. I then lowered Ch1’s R56 very slightly to get the compression symmetry of that last drum hit and cymbal decay to match visually on the Pro Tools meters and to my ears. With the multed tracks panned L and R it now sounded like a mono track down the middle, with no more L to R back and forth warble. After that I double checked with a low to high volume (from no compression to -16db) sweep of 1k and it looked and sounded good. I didn’t run it on a graph to check for perfection, but it sounds very right to me. The way the THD ended up for me is plenty enough of it at 12 o’clock input with program material, and 2-3 o’clock with a single track of drums or guitar, and output set for unity gain between Bypass and THD. I hope this helps!

I should also mention that I matched every resistor exact (to whatever decimal as far as my dmm would read) between the two channels, including the attack switches. I also matched the caps on the sc hpf.

 

[thelivingroom]

Not a stupid question! Yes the corresponding pairs on each channel are matched to 0.01V.

Not sure if this made the difference on mine or not, but I matched all 8, not just the pairs, to .001v at both 1k and 10k.

 

[SmokeyJoe]

First I did Fine Tuning Step 1...

Thanks for the detailed response, that's super helpful.

Not sure if this made the difference on mine or not, but I matched all 8, not just the pairs, to .001v at both 1k and 10k.

I wonder if that's it. I matched D1/D3 and D2/D4 between channels, but I didn't match all 8. I figured that would be enough for symmetry between channels.

Earlier in the thread fripholm mentioned that 1% deviation (+-45mV) was more than enough accuracy for the diodes, so I figured my measurements were accurate enough. Maybe that's worth trying though.

I didn't match components between channels other than the input/output rotary switch resistors, but I did use 1% metal films everywhere. I'm happy for things not to match perfectly but my curves are way out.

Just had another look at your build, so much attention to detail! What enclosure did you use? Those mounting rails are neat.

 

[thelivingroom]

Thanks for the detailed response, that's super helpful.


I wonder if that's it. I matched D1/D3 and D2/D4 between channels, but I didn't match all 8. I figured that would be enough for symmetry between channels.

Earlier in the thread fripholm mentioned that 1% deviation (+-45mV) was more than enough accuracy for the diodes, so I figured my measurements were accurate enough. Maybe that's worth trying though.

I didn't match components between channels other than the input/output rotary switch resistors, but I did use 1% metal films everywhere. I'm happy for things not to match perfectly but my curves are way out.

Just had another look at your build, so much attention to detail! What enclosure did you use? Those mounting rails are neat.

I’m sure your diodes are matched well enough, and I don’t know how much of a difference all the matching did on mine vs. sticking with the 1%. I tend to go overboard when I have plenty of time between saving up for the next phase of orders or waiting for backorders/shipping from other countries haha.

Thank you! The case is from Don Audio. I usually use Collective Cases for simpler builds because they look nice and are here in the U.S., but I was already ordering the meters and knobs from Don and liked the idea of the rails and being able to do the extensive amount of wiring and mounting of things with the top and bottom panels removed. It made it a lot easier for sure. It’s a great case!

 

[TwentyTrees]

It’s a great case!

They're also available from Frank Rollen at frontpanels.de as his "NRG" case line, and I fully agree - they really are great. Frank carries a bunch of accessories for them as well, things like stabilisation brackets, toroid mounting plates, etc. I use them for all my builds.

 

[thelivingroom]

They're also available from Frank Rollen at frontpanels.de as his "NRG" case line, and I fully agree - they really are great. Frank carries a bunch of accessories for them as well, things like stabilisation brackets, toroid mounting plates, etc. I use them for all my builds.

I saw before that they looked identical and he has even more accessories for them than what’s on Don’s site. One of these days I’ll get a case and front panel from him. I’ve heard nothing but great things about the front panels for years.

 

[SmokeyJoe]

After a lot of testing I'm pretty sure I've narrowed the issue down to the CVs not matching, causing one channel to compress at a higher ratio. I've used FT2B to (approximately) match the output levels up to the threshold (the CVs match up until this point), and beyond that they start to diverge. With a 0dBFS 1kHz tone, the CV measures 3.98v on the left channel and 3.35v on the right, resulting in the right channel being ~2.4dB louder. With no signal the CVs are at 5.65v for left and 5.52v for right.

- I've tested every trimmer in multiple positions throughout their range
- I've swapped the boards and the problem moves with the board
- injecting signal at byp2 middle pin results in the same level output on each channel, so the issue is before that
- I've measured some discrepancies in voltages between channels at the collector of T8 - 1.72v vs 2.04v
- my 20v rails differ slightly at 19.86v vs 19.93v, not sure if this is enough to cause an issue but I could more closely match R31
- behaviour is similar in comp mode and limit mode

I'd appreciate any suggestions as I'm pretty much at the limit of my ability to fix this. All I can think to do next is start swapping out transistors and diodes until things start working.

 

[thelivingroom]

A little update on mine: I changed the input pots from linear to log, and am much happier with the range at the beginning of the rotation now.

 

[SmokeyJoe]

Update: Maybe getting closer to the problem. I've been probing a sine wave with my scope - in the attached image, anywhere marked with a blue star is where the signal level was lower on channel 1. The issue starts at T1 and T2, but it looks like they both receive the signal independently so seems unlikely they are both faulty.

What's odd is on channel 1, I can trace the signal from the input up to D2 and D4, but it disappears after D2/D4 and before D1 and D3 (marked with blue lines). On channel 2, I can trace it past both sets of diodes as far as R14 and R15. I'm not sure what this means but it seems significant, maybe someone is able to shed some light?

I'm beginning to wonder if the zeners are somehow at fault, perhaps I do need to match them more accurately.

Thanks.

 

[thelivingroom]

Update: Maybe getting closer to the problem. I've been probing a sine wave with my scope - in the attached image, anywhere marked with a blue star is where the signal level was lower on channel 1. The issue starts at T1 and T2, but it looks like they both receive the signal independently so seems unlikely they are both faulty.

What's odd is on channel 1, I can trace the signal from the input up to D2 and D4, but it disappears after D2/D4 and before D1 and D3 (marked with blue lines). On channel 2, I can trace it past both sets of diodes as far as R14 and R15. I'm not sure what this means but it seems significant, maybe someone is able to shed some light?

I'm beginning to wonder if the zeners are somehow at fault, perhaps I do need to match them more accurately.

Thanks.

Just brainstorming a couple things, as I would like to help but I’m also not mega experienced with all this myself. What values are FT2B set to? When set to higher than a certain R with a 0db tone, mine would also partially cut out. Have you swapped just the Zeners, and if so does the issue of losing signal after D2/D4 follow the swap? Since there are several resistors around all that, I’d also double check them for correct value. Are you measuring the voltage at the input pot wiper when calibrating to make sure both channels are receiving the exact same level?

 

[SmokeyJoe]

Finally found the issue. I feel annoyed with myself for missing it but mostly just relieved to have it sorted.

After figuring out the signal stopped after D2/D4, I realised the diodes were shorted at that point, and a look with my loupe confirmed a tiny solder bridge from when I'd swapped out some noisy diodes. After removing that, the attached graph shows the response with default trim settings is very close, and I'm confident I can get them to match after further calibration.

Thanks to all those who offered suggestions.

Just brainstorming a couple things

Appreciate the suggestions - swapping the diodes was going to be my next thing to try (and likely would have fixed it, but probably not for the reason I would have thought!)

 

[TwentyTrees]

Great, glad you got it sorted!

 

[thelivingroom]

Finally found the issue. I feel annoyed with myself for missing it but mostly just relieved to have it sorted.

After figuring out the signal stopped after D2/D4, I realised the diodes were shorted at that point, and a look with my loupe confirmed a tiny solder bridge from when I'd swapped out some noisy diodes. After removing that, the attached graph shows the response with default trim settings is very close, and I'm confident I can get them to match after further calibration.

Thanks to all those who offered suggestions.


Appreciate the suggestions - swapping the diodes was going to be my next thing to try (and likely would have fixed it, but probably not for the reason I would have thought!)


View attachment 109831

Awesome! Great job; I’m happy you got it sorted! Often times it’s something so simple that can set us back days. I know the feeling!

 

[Hobbyist]

Hi all,
Finally finished my unit as well. Sorry my english is not super technical so I’ll try to explain my build history and debugging process as clear I can.
The case was quite small for this project so I tried to cage everything as much possible to avoid unwanted radiation from the Toroidal transformer. I've learned a lot from this project through the years as this system of compressors is totally new for me.

Matching PCBs:
I’ve matched almost every component from pcb L and R to 1% accuracy. Aswel for the sidechain and the Elmas type 3 switches are all matched to 1%. I’ve also placed a trimmer on the output to match the Carnhills, i've noticed when just letting the signal going through the iron they have almost 1db stereo tolerance .. is that normal for these expensive irons ?
only with the trimmers after the output stage I'm able so far to match the unity gain properly.

Zener Diodes:
The diodes I’ve bought 4 years ago are matched too (forgot which brand). the noise is around -114 on both sides which is quite ok for me. But there's hum and i don't know what's causing this?

Bypass switch pcbs:
The cables for the bypass are very very sensitive so tried to shield them to test out if it was working and surprisingly it worked well to move on. should the 3 cables on the bypass insert on the main pcb be more separated?

Hum:
Like I stated before, There is hum going trough both channels. Channel R haves more hum than channel L as you can see on the Pics. Channel L is quite acceptable. I’ve tried to figure out what the problem is causing the bigger hum on channel R. Anybody experienced this as well? is it caused by the diodes? just to understand what's going on here.

Channel L


Channel R


Grounding:
I’ve grounded everything according to this block diagram.


Input Transformer:
From the input transformer VTB 9046M
Pin 6 is going to Signal ground.
Pin 11 is going to Chassis ground.

I can switch the primarie in parallel to boost the input gain +6db. Don't know if I understood well from the Build Guide, 2:1 can bring damage to the output stage? If i put heatsinks on the output transistors, will it be ok?

PSU:
The second PSU is regulated to 12V and its used only for the relays and leds. tapped from 0V and +28V from the main PSU.

Tuning:
This was an intensive and time consuming task. i’m lucky here in this thread a lot of info is available and helped me a lot trough the final steps. thank you @thelivingroom, @SmokeyJoe and many others for sharing your experiences. If anyone can give me hints to get rid of the hum would be much appreciated!

Thank you @fripholm for the amazing pcbs and ideas. I’m super happy with the sound, it was worth waiting for it!

All the best,
Michael

Some more pics:

 

[SmokeyJoe]

There is hum going trough both channels. Channel R haves more hum than channel L as you can see on the Pics. Channel L is quite acceptable. I’ve tried to figure out what the problem is causing the bigger hum on channel R. Anybody experienced this as well? is it caused by the diodes? just to understand what's going on here.

Looks like a nice build, congrats. Sorry you're having trouble. As far as I'm aware, the faulty diodes don't cause hum - it's more of a pink noise. That's how my first set sounded.

A few suggestions come to mind:

- Does the hum change with input or output gain? Or with compression?

- It looks like you've connected grounds to the chassis in several places. Try chaining your XLR grounds together and bringing everything to a single point on the chassis, along with the earth and PSU ground. See this post by fripholm for examples. It looks like you also have a ground connection on the front panel - what is the purpose of that?

- Try using shielded cables between both sets of XLRs and the transformers. That's the only place I've done so - my bypass cables are neither shielded nor twisted.

- Try experimenting with where you connect pin 6 and 11 of the input transformers.

- Double check all of your grounds are making good contact.

- Test with the lid on the enclosure, in case it helps with blocking interference.

- Try in a different socket, ideally on a different ring. You could even try in a different building.

It feels odd to me that there would be a 1dB mismatch in your transformers, mine are very close - less than 0.1dB iirc. Can you work out whether it's the inputs or outputs?

Your noise level of -114 seems remarkable - what scale is that? Is that at unity gain?

 

[Hobbyist]

Looks like a nice build, congrats. Sorry you're having trouble. As far as I'm aware, the faulty diodes don't cause hum - it's more of a pink noise. That's how my first set sounded.

A few suggestions come to mind:

- Does the hum change with input or output gain? Or with compression?

- It looks like you've connected grounds to the chassis in several places. Try chaining your XLR grounds together and bringing everything to a single point on the chassis, along with the earth and PSU ground. See this post by fripholm for examples. It looks like you also have a ground connection on the front panel - what is the purpose of that?

- Try using shielded cables between both sets of XLRs and the transformers. That's the only place I've done so - my bypass cables are neither shielded nor twisted.

- Try experimenting with where you connect pin 6 and 11 of the input transformers.

- Double check all of your grounds are making good contact.

- Test with the lid on the enclosure, in case it helps with blocking interference.

- Try in a different socket, ideally on a different ring. You could even try in a different building.

It feels odd to me that there would be a 1dB mismatch in your transformers, mine are very close - less than 0.1dB iirc. Can you work out whether it's the inputs or outputs?

Your noise level of -114 seems remarkable - what scale is that? Is that at unity gain?

Thank you for the information! Lot to learn for me here.
-The Hum doesn't change when changing input or output gain. when in bypass or compression in, the hum is always there. if i disconnect the Right output XLR, You can hear clear on the left channel the hum is much less annoying. It's just the right channel having this problem.
-Will fix the grounding! I've made a ground on the frontpanel when I was testing with the panel open. I'll remove that one too.
-The -114 noise is when the compression is in or out. small differences between the two.
-I have to double check with the input or output transformer differences. maybe a bad solder or cables like you mentioned.

Guess I have to look for some new cables and a new bypass system.

So many things i can change and test with your information. will try out when i've time and keep posted. Thanks again !

Scale is Db non linear. with no audio and compression IN, volumes at unity gain same as bypass. Here is another picture with both channels:

 

[SmokeyJoe]

Thank you for the information! Lot to learn for me here.

No problem. And don't worry - if you've made it this far there's nothing in there that will be beyond you, just be methodical.

I have a hunch the issue is somewhere in the right channel after the output gain pot, if that's not affecting the hum. I would start with using the suggested grounding scheme and see if that helps, followed by shielding your cables to the XLRs, as these are both recommended by fripholm.

Good luck!

 

[TwentyTrees]

Your output transformers are unshielded, and are quite close to your power transformer. Does the hum change in nature if you rotate the latter? (Obviously disconnect power, or use rubber gloves for this bit!)

Toroidal power transformers can induce hum in other magnetics. Shielding helps; distance helps more; and the exact angle of the toroid can make a big difference, as the part where the wires emerge tends to be the "noisiest" angle in my experience. Often just rotating it and measuring the result can get you a good outcome.

Good luck!

 

[thelivingroom]

Hi all,
Finally finished my unit as well. Sorry my english is not super technical so I’ll try to explain my build history and debugging process as clear I can.
The case was quite small for this project so I tried to cage everything as much possible to avoid unwanted radiation from the Toroidal transformer. I've learned a lot from this project through the years as this system of compressors is totally new for me.

Matching PCBs:
I’ve matched almost every component from pcb L and R to 1% accuracy. Aswel for the sidechain and the Elmas type 3 switches are all matched to 1%. I’ve also placed a trimmer on the output to match the Carnhills, i've noticed when just letting the signal going through the iron they have almost 1db stereo tolerance .. is that normal for these expensive irons ?
only with the trimmers after the output stage I'm able so far to match the unity gain properly.

Zener Diodes:
The diodes I’ve bought 4 years ago are matched too (forgot which brand). the noise is around -114 on both sides which is quite ok for me. But there's hum and i don't know what's causing this?

Bypass switch pcbs:
The cables for the bypass are very very sensitive so tried to shield them to test out if it was working and surprisingly it worked well to move on. should the 3 cables on the bypass insert on the main pcb be more separated?

Hum:
Like I stated before, There is hum going trough both channels. Channel R haves more hum than channel L as you can see on the Pics. Channel L is quite acceptable. I’ve tried to figure out what the problem is causing the bigger hum on channel R. Anybody experienced this as well? is it caused by the diodes? just to understand what's going on here.

Channel L
View attachment 109882

Channel R
View attachment 109883

Grounding:
I’ve grounded everything according to this block diagram.
View attachment 109884

Input Transformer:
From the input transformer VTB 9046M
Pin 6 is going to Signal ground.
Pin 11 is going to Chassis ground.

I can switch the primarie in parallel to boost the input gain +6db. Don't know if I understood well from the Build Guide, 2:1 can bring damage to the output stage? If i put heatsinks on the output transistors, will it be ok?

PSU:
The second PSU is regulated to 12V and its used only for the relays and leds. tapped from 0V and +28V from the main PSU.

Tuning:
This was an intensive and time consuming task. i’m lucky here in this thread a lot of info is available and helped me a lot trough the final steps. thank you @thelivingroom, @SmokeyJoe and many others for sharing your experiences. If anyone can give me hints to get rid of the hum would be much appreciated!

Thank you @fripholm for the amazing pcbs and ideas. I’m super happy with the sound, it was worth waiting for it!

All the best,
Michael

Some more pics:
View attachment 109885View attachment 109886View attachment 109887View attachment 109888View attachment 109889

It looks very nice! A few things could help with the hum and overall layout:

1) I would have used 24v relays and eliminated the extra wiring and boards involving having the 12v psu. They can usually handle 150% of the rating, so 28v is totally fine. That would clear out a good amount of space. I daisy-chained +28v to each relay +, and switch the - to actuate them. No pops whatsoever when I switch bypass, link, or auto release.

2) I would use shielded cable for I/O XLR’s and transformers to boards, as well as I/O gain switches. I left the output XLR shield disconnected at the output trans side so no loop is possible, and it drains at the star ground via Pin 1 of the XLR.

3) My grounds tie to one star chassis point only: Pin 1 of I/O XLR’s, mains earth, and psu output (6 total). Input trans secondary signal cable - (Pin 10) and shield (Pin 6) go to the output - on the pcb, and I made sure to not tie any L or R channel pcb signal grounds together until they meet at the psu out ground via STROM. I probably spent the most time carefully planning all of that out, based on the information @SmokeyJoe linked above.

I hope this helps! If I think of something else I’ll report with more.

 

[Hobbyist]

Your output transformers are unshielded, and are quite close to your power transformer. Does the hum change in nature if you rotate the latter? (Obviously disconnect power, or use rubber gloves for this bit!)

Toroidal power transformers can induce hum in other magnetics. Shielding helps; distance helps more; and the exact angle of the toroid can make a big difference, as the part where the wires emerge tends to be the "noisiest" angle in my experience. Often just rotating it and measuring the result can get you a good outcome.

Good luck!

You're right, I was afraid for this as well, the shielding I did already had a good result, maybe it's not good enough. will do some test.
And maybe the PSU is too close to the front panel - maybe too close to the Input and output pots from the right Channel where the problem is most noticeable .

Some work and tests to do the next days.

If I could start it all over, I'll go for another enclosure and design - this one is a pain to work with.

Thank you for the advice !