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

[Potato Cakes]

If you increase the input level, you get higher level across the zeners. This will get you higher THD. You can increase the level only so much before you get distortion of unacceptable levels.

This is one of things I thought about and was afraid of happening. Maybe R14 and 15 could also be tweaked so as to not prematurely overload the zeners. I'll poke around with this after I swap zeners and see if I get better noise measurements.

Thanks!

Paul

 

[fripholm]

As Ilya said, in this circuit, noise behavior is extremely dependent on distortion behavior - there's no free lunch. When I designed the boards, one of the goals was to lower the noise floor. I actually modified the level going into the sidechain - which in turn altered the knee of the compressor and the ratio, which also needed to be compensated and so on... 

This is a feedback topology, where everything depends on anything else. Changing the overall behavior is not as trivial as it seems. Whoever the genius was, who designed the original, I'd bow to him

 

[Potato Cakes]

As Ilya said, in this circuit, noise behavior is extremely dependent on distortion behavior - there's no free lunch. When I designed the boards, one of the goals was to lower the noise floor. I actually modified the level going into the sidechain - which in turn altered the knee of the compressor and the ratio, which also needed to be compensated and so on... 

This is a feedback topology, where everything depends on anything else. Changing the overall behavior is not as trivial as it seems. Whoever the genius was, who designed the original, I'd bow to him

Copy that.

Hopefully the new zeners will give me what I'm after, or at least get me closer.

Thanks!

Paul

 

[schrobbelbop]

Hi Guys,

maybe not for this project but i'm building diy-racked's version of the zener limiter.
it's not an easy beast to manage!

i have one issue left, when linked the left channel is giving more compression then right above 6db's of compression
till 6db's it stays fine with right.

left channel CV raises at link point where the right channel stays the same.
TP-E goes down around 1.5 vdc on the left channel.

replaced and removed D-1-4
matched Q13/14

i have replaced 1n751 zeners from mousers brand to farnells (multicomp) the mouser ones are unusable (brand: Central) lotsa noise on the output.
could also be the other zeners be totally diffrent? it seems to be that when above a certain amount off compression the knee gets very steep and it turns in to infinity limiting....

 

[Potato Cakes]

Hi Guys,

maybe not for this project but i'm building diy-racked's version of the zener limiter.
it's not an easy beast to manage!

i have one issue left, when linked the left channel is giving more compression then right above 6db's of compression
till 6db's it stays fine with right.

left channel CV raises at link point where the right channel stays the same.
TP-E goes down around 1.5 vdc on the left channel.

replaced and removed D-1-4
matched Q13/14

i have replaced 1n751 zeners from mousers brand to farnells (multicomp) the mouser ones are unusable (brand: Central) lotsa noise on the output.
could also be the other zeners be totally diffrent? it seems to be that when above a certain amount off compression the knee gets very steep and it turns in to infinity limiting....

You are right, that is not for this project. It is best you email them directly.

Thanks!

Paul

 

[Olegarich]

Are those boards still available? wrote you a mail today.

 

[fripholm]

Are those boards still available? wrote you a mail today.

Sorry, I didn't get any mail from you. The boards are available!

 

[Potato Cakes]

I'm home for a minute so I got a chance to swap out the 1N751As with 1N5231s and the noise is drastically lower. Compression and limiting seem to behave and sound as before, but i think I might have to drive it a little harder than before to get the same gain reduction. I'm not entirely sure on that one as it's been a while since I've tinkered with this. Now I just need to finish matching distortion and level between the channels and I should be good to go.

Thanks!

Paul

 

[Potato Cakes]

Now both sides seem to be matched and the noise floor is down where I can finish tidying up the wiring inside and put it to use. You can see that I was able to get the 1kHz level and the third harmonic to match up pretty much perfectly.  I definitely didn't expect to have to experiment and solder/desolder as many times as I did, but I managed to learn a couple of things in the process. I still haven't been able to get rid of all the popping when using the switches, and maybe some day I'll get back to that. Here are some final notes regarding building these boards....

First, the circuit works exactly as is when built per the build manual. It seems to work well with the 2.4mA meters that Don Audio sells for this type of zener limiter.

Stereo matching channels is very tricky. R22 does indeed affect distortion, but so does R56 and the meter trimmer. This is also compounded when trying to match levels with R56. And you have to take into consideration that this same R56 value will affect the meter position, so when switching to compress/THD modes your metering will not be the same. For me, one of the channel's meters is slightly higher than the other in resting position after doing all of my tweaks, but the tracking above 3dB of reduction is very close. I think this board would be well served with an optional trimmer in the output stage to match levels without affecting the other aspects of the circuit. I think this would make matching much easier.

Also for the distortion calibrating, it's best to start with the attack set all the way to fast and not on auto release if you built your unit with this (I had previously stated to have the auto release engaged, but my push/pull switch was wired backwards). When you adjust R56 to give you maximum output, there is about a 1dB difference in level when the release control is adjusted as well as affecting the amount and type of distortion. However, when R56 is adjusted to match one channel to another, then the amount of output and distortion affected by the release control is quite a bit more, so it becomes necessary to have both release controls in the identical position when starting this process. I found that fully CCW (fast) is the best place to start so you have the most distortion present that can be seen on a meter as it relates to the release control. After all the tweaks are made, you can then used this control to change the distortion characteristics.

I used all stepped controls except for the release, but I feel that maybe a variable control for the output would be better when trying to use this for stereo applications and one is trying to match left and right. Even with using 1% resistors, the output levels are a bit off when both controls are set the same. Again, this may also be solved with a trimmer in the output section. This can be easily compensated in the DAW, but if you are unnecessarily obsessed with perfection when calibrating builds the slight mismatch can be off-putting.

Lastly, I don't think one should be focused on making a perfectly matched stereo unit with this circuit. I can't imagine the originals were, especially with component tolerances back then. One could match all the resistors, diodes, and transistors to 0.1% or better, but I honestly don't think it would be a good use of one's time for a dynamic processor that isn't known for precision and finesse. That's just my opinion, though.

Thanks!

Paul

 

[Potato Cakes]

Tinkered with this just a little bit more before putting the lid back on for good, which gave me some new results...

I put C34 back in the one channel that was still popping (the other one I hadn't removed). Now I can see meters jumping on the inputs of my DAW when I take it in and out of circuit, but I cannot hear it whereas before it was a pretty massive sound. Not sure what I did, but somewhere in the 100+ times I took the boards apart and I managed to "fix" it.

The other thing is the linking. I had spent so much time tweaking the individual channels to match, but when I engaged the link there was about a 3dB loss on one side when just using tone. However, this is not necessarily the case when actually running program through it. I had this similar discussion with Mike at Hairball Audio when using the linking cards on his FET rack compressors and he concurred that tone and program yielded different results. So one will have to lean more on listening skills rather than having eyes glued to a set of meters, if one can imagine such a thing.

I also wound up putting both meters back to where they were for the initial calibration. After the unit being powered on for some time the resting position changed. This "warm up" time seems similar to the ez760 calibration.

If there were any future revisions, I would recommend having the option to permanently install a trimmer for R22 and a trimmer in the output section for level matching. I think with those changes R56 could just be 1k as the BOM states.

I would be interested if someone here did some transformer testing to see if there were any notable changes in audio characteristics. But emphasis on someone else doing that. I believe I have had my fill experimenting with this board.

Hopefully all of my struggles (mostly self-induced) as documented on this thread will help others who are building this circuit to complete one of these boards faster and without headaches.

Thanks!

Paul

 

[fripholm]

Great recap, Paul! Thanks for your input and persistence - much appreciated. 

Do you mind if I'd take some of your findings about the calibration procedure and add it to the guide?

 

[Potato Cakes]

Great recap, Paul! Thanks for your input and persistence - much appreciated. 

Do you mind if I'd take some of your findings about the calibration procedure and add it to the guide?

Absolutely use any of this information that you want. I would recommend double checking my findings as I finished this up really late last night, but I did sit with this for a while doing tests with program material, so I'm confident that these results are repeatable. It might be helpful in the future to have some tips on using this in stereo operation, but I feel that if one acquires enough DIY experience to successfully build this circuit then they should be able to take into account the factors affecting the stereo linking operation.

I'd love to hear from someone who has built one of these with the hold control. I'm curious if this option would be more useful than the auto release. I have plans on building a channel strip with this as the compressor section and am thinking about going this route for the next one.

Thanks for making this project available and taking the time to provide updated information. I thought about renting a Chandler unit and making comparisons, but if someone wants a Chandler then buy one. Based off of my recollection, this unit behaves very similar, plus you get to learn stuff in the process.

Thanks!

Paul

 

[Ilya]

Having built this compressor per original schematic I can comment on hold control. I rarely leave it at maximum as in Chandler. Most of the time it is somewhere between 2/3 and 3/4 rotation where compression kicks in, but is not crushing the signal too much.
I had no difficulty matching channels. I had 4 close pairs of zeners though. I get a good match in sonic signature and THD, and tracking is very good both with link engaged and in independent operation. When linked you have to set all controls identically (except output) otherwise the units will not track properly.
The real task is finding proper zeners and matching them. Everything else is not that difficult.

 

[shot]

Paul,
Thank you for doing this in such detail. It's so inspiring to see someone go through such struggle and document every step with successful end.
I feel now that I have to find spare time to sit and finish my two channel unit! It got stuck on the bench for months since I'm too tired when I get home late and I had no energy to do my diy stuff.

Luka

 

[shot]

Well, I had some time to play with my dual unit.
Seems that the right channel is working fine, except for some 50hz hum that is due to position of the output transformer. I'll move both transformers to a location that is not so close to psu and it's power transformer.

But I have problem with left channel.
Before I start fiddling with trimmers, I wanted to check if anyone have any idea where to look for a fault.
Audio is passing fine and it compresses (and limits) under some conditions. When the release is from 12 o'clock up to CW position it's okay. Compression is happening (though I seem to have a bit more distortion combined with some hiss whenever compression is going on, but I think it's due to trimmers being untouched). But when I shorten the release, at some point it just slams the gain reduction to max and output level drops about 20db down. And it stays like that until I move the release to some longer value past the 12 o'clock. It seems it doesn't have anything to do with attack time, it's just release related.

Any ideas where to look for a problem?


Luka

 

[fripholm]

Might be that the control voltage makes it through as DC into your audio signal. When it does that, does your audio look like on the picture? If so, you might need to do Fine Tuning 1 as described in the guide.

 

[shot]

Fripholm, I went to see what the waveform looks like and it's not raised up by DC signal.  I tried to adjust R7 but I couldn't manage to get my signal to have flats. Right channel was easy to adjust, I did that already, but on left channel it was impossible. I think it has to do with my main problem (see below)

I sorted out my initial problem with release cutting the signal if set too short! Left channel was way too low in level. So I went to increase R22 and when I got it to be the same level as right channel, all the problems with release were solved. I fiddled a bit with R22/R56 and settled on some setting that is very similar on both channels.

But I still have an issue with left channel!
Whenever there is gain reduction, there's a lot if hiss added to the signal. It happens only when there is compression going on, so it's not something with output amp stage. Also, maybe related to the same problem, when I switch to THD setting there's a lot more upper harmonics. It's like ten upper harmonics present above 1khz test tone. Regulating R22/R56 doesn't move them, just the first three or four harmonics move (as they should).
I'm thinking it's got something to do with zener diodes. Should I try to replace them? Or is it possible that it's some faulty transistor?
I'm tempted to order 1N5231 and put those on both channels since Paul reported that they lowered the noise compared to 1N751A.

Btw, tonight was the first time I was able to listen to this compressor! Even though it's still not finished, it's already KILLER sounding! I've built PYE, PRR176, Pico, clx160,  few 1176s and this one has something special in it's sound. I'm already happy with it! Can't wait to fix this hiss issue, add hold control and take it to the studio!



Luka

 

[Potato Cakes]

Did you see my above post regarding THD? I was getting more than ten harmonics per channel. Where do you have your release control when you started matching THD? They must be in the same exact position when you start messing with R22 and R56. Plus, you really need to make sure when using trimmers for R22 and R56 that they are adjusted to the same value when starting to do any sort of matching. Remember that R56 affects the harmonic content as well as R22. It's a bit of a balancing act until you arrive at a result you like.

The hiss doesn't make any sense. Does it hiss when it is not compressing? When I had a noise issue it was ducked down when it was in gain reduction. I believe if the zeners were the issue the hiss would be present even with no signal passing through.

Thanks!

Paul

 

[shot]

Did you see my above post regarding THD? I was getting more than ten harmonics per channel. Where do you have your release control when you started matching THD? They must be in the same exact position when you start messing with R22 and R56. Plus, you really need to make sure when using trimmers for R22 and R56 that they are adjusted to the same value when starting to do any sort of matching. Remember that R56 affects the harmonic content as well as R22. It's a bit of a balancing act until you arrive at a result you like.

Paul,
Yes, I've set trimmers on both channels to the same position at first. But my first channel was way off at that point. It was ducking the signal if release was set shorter than 12 o'clock. So I figured that it is due to R7/R8 and when I increased it enough the release problem was solved. During that process I had to trim R22 and R56 to get that channel's distortion and level closer to second channel. And during that tuning I noticed that I can't get rid of higher harmonics no matter what. And hiss issue was there all the time.

The hiss doesn't make any sense. Does it hiss when it is not compressing? When I had a noise issue it was ducked down when it was in gain reduction. I believe if the zeners were the issue the hiss would be present even with no signal passing through.

Thanks!

Paul

It's interesting - it doesn't hiss at all when there's no compression! It's just there when compression kicks in, and hiss level doesn't ride in level with amount of gain reduction. That's how I assumed it may have something to do with zeners... or not...?
And again, this happens only on the first channel! I went to measure some voltages on transistors and compare them channel to channel. It's very consistent. May be few mV off here and there but it all seems normal. So now I have no idea what to do apart from ordering new zeners and trying them. But still why would I do that if my second channel is okay, and all the components used throughout the build were from the same stock and I tried to match them whenever I could.
I have no ideas...

Luka

 

[fripholm]

I've just got a report back from another guy who also used the zeners from Mouser. His stereo unit developed a massive hiss when turning the release pot clockwise. He's now replaced those with 1N751s from Farnell (Multicomp) and the hiss is gone and the unit's working as expected.

It seems, most of the noise problems arouse from those Mouser zeners (Central Semiconductor?), although the symptoms may vary a lot...