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

[brewbacca]

I haven't tried the resistor yet, but I did seem to eliminate the static by redoing the wiring using shielded cable. I think this also helped cut down on the hum some more. I definitely get a hi-pitched oscillation when I unplug the byp2 connection, but since that won't be the case when I'm using the unit, I don't think it's anything to worry about.

Now I'm getting an issue where the second channel is really quiet, but I'm assuming I messed something up doing the rewiring, so I get to check my work again. I think this thing is cursed!

 

[octavez]

I'm also having what I supposed are diode noise issue, but before ordering a bunch more here is an audio test so you guys can confirm it (the drums loop stops at 23s and the noise will rise up):

https://www.dropbox.com/s/seqjvnnj87pr4gc/zener-limiter-noise-1.aif?dl=0

The diodes I'm using are from Mouser: On Semi 1N5231C

Thank you!

 

[fripholm]

This definitely sounds like a lot of compression on the drums before the noise kicks in. Was this recorded with the level controls fully clockwise?

If I crank the input and output controls all the way up on my units, the noise is almost that high BUT it's pure white noise whereas yours is more pink-ish with lots of low frequency movement and a rustling paper-like quality to it. I haven't heard that before. The compression sounds great, though.

As I was lucky enough to have not run into noisy Zeners myself, I can't really tell whether the diodes are at fault here but it is possible. Maybe the noise sounds familiar to someone else who had to deal with this problem...

BTW: these are the ones from my units. HTH

 

[octavez]

Thank you for the update! So with no signal that crackly pink noise is constant at around -40db wherever the setting is on the input attenuator (output attenuator all the way up), that sounds a bit too much right?

 

[fripholm]

at around -40db...

I don't know what those 40dB are referenced to (0 dBu, dbV, dBFS etc.) but generally speaking, if there's just 40dB of headroom then the noise level is too high, yes.

 

[octavez]

I don't know what those 40dB are referenced to (0 dBu, dbV, dBFS etc.) but generally speaking, if there's just 40dB of headroom then the noise level is too high, yes.

I meant -40dBFS.

 

[octavez]

I managed to find some nos Fairchild 1N751A. The unit is dead silent now!! So yeah those 4 diodes are a pretty big deal.

I was wondering if there is a proper way of measuring the amount compression, to make sure the meter is properly calibrated?

 

[fripholm]

The build guide for my OP76 FET compressor boards explains a very similar procedure where you're using your DAWs level meters to calibrate the gain reduction meter. Both compressors share the same workflow/mindset with their fixed threshold so it should be easy to adapt.

You can find it in the first post of the OP76 build thread here. It's on page 7 under "Gain reduction meter tracking".

You're basically looking for a fixed difference between input and processed (reduced) output level and adjust the GR meter to that value. Don't expect high precision.

Sorry for the shameless advertising  8)

 

[octavez]

Haha awesome will check it out! Thanks a lot for being so responsive!

 

[JMan]

After putting it off for awhile, I'm returning to finish up this project.  I've created a complete wiring diagram for myself, but I'm wondering about this connection.  I've never done a stepped attenuator before.  I'm following the build manual for resistor values on a 24-position switch -- I splurged on Elmas -- but I want to double check that my wiring connections are correct.  Can someone confirm if I've done this right or if I've got it backwards?  (I'm attaching an image of the input connection, but my output looks the same -- square pad to pos 24/full CW, middle pad to pole, third pad to pos 1/full CCW)

 

[fripholm]

Had to open my stereo unit to confirm: your diagram is correct.

Might help others as well, so thanks for that!

 

[JMan]

I spent the evening soldering the input, output, and attack switches.  Glad to have them done!  The outs just about made me cry, as I got them fully completed and then realized I had done them the wrong way round AND had screwed up my start position by one lug — meaning I had to remove ALL of the resistors on both switches and completely redo them!  :'(  They aren’t as pretty as the input switches anymore, but they’re done!  I matched resistors between the switches to the smallest decimal on my Fluke 179, so they measure very tightly.

 

[octavez]

If I want the Elma input and output attenuators to be 22 positions instead of 24 (like the original). Which steps from the guide should I skip? On the output attenuator the markings indicate 0db at noon (so halfway).

Thank you!

 

[JMan]

Correct me if I'm wrong, but I think this is purely intended to be an attenuator -- as in, no "+" side of things, but rather providing full signal at the furthest clockwise position and stepped attenuation as you move CCW. 

I see what you're looking at on the Chandler version.  I don't know enough to say if that is providing actual gain in hand (as it seems to imply) or not.

 

[octavez]

Correct me if I'm wrong, but I think this is purely intended to be an attenuator -- as in, no "+" side of things, but rather providing full signal at the furthest clockwise position and stepped attenuation as you move CCW. 

I see what you're looking at on the Chandler version.  I don't know enough to say if that is providing actual gain in hand (as it seems to imply) or not.

I'm wondering if it is supposed to be gain unity when output attenuator is at noon and input attenuator all the up (no attenuation). The original EMI module as the same -10db/+10db markings:

 

[JMan]

I'm wondering if it is supposed to be gain unity when output attenuator is at noon and input attenuator all the up (no attenuation). The original EMI module as the same -10db/+10db markings:

Ah, got it.  Interesting, I'm not sure if unity is meant to sit at 12 o'clock or not.  Guess we'll have to wait for Fripholm to weigh in.

One thing that does occur to me is that, IF unity is indeed meant to be at noon, AND you want to reduce the number of positions on your attenuator switch, you'll probably have to do some re-calculating of resistor values on the switch to make that all line up accurately (as opposed to simply skipping certain steps).  In other words, imagine that with 24 positions at 1.5dB steps you hypothetically arrived at unity after 11 positions (I'm not sure if this would be exactly noon, but it doesn't matter) -- but you want to use 22 positions, so in order for it to still work and look symmetrical on the frontpanel control, you now need to get to unity in 10 steps, meaning you'll have to change the step sizes to something like 1.65dB each, meaning all those resistor values would need to be adjusted.  I've no idea if this sort of precision is possible given standard resistor values.  I also might be missing something obvious, but it seems to me like this combination of goals would require some thought.

 

[fripholm]

The input and output controls are pure attenuators and the output and compressor stage each add some fixed gain which also depends on the calibration with the latter.

I don't know whether Chandler has a standardized calibration procedure for their units to achieve unity gain at certain knob positions - I certainly have not. With a design like this (fixed threshold that is), the output level is always a moving target as you never know how much compression leads to whatever output level. You'll usually adjust compression amount by ear and make up the output level accordingly to achieve unity gain.

In this case certain values around the knobs wouldn't make much sense anyway - apart from being able to recall specific settings later. But these values can be anything really - like a scale from '1 to 24' or '-100 to +100' or whatever. It's music not math As mentioned in the guide, the step sizes I came up with are pretty odd but they work fine in practice. I used LTSpice and a lot of trial and error to get there  8)

FWIW, unity gain is at 11 o'clock on my stereo unit for both the input and output knobs with no compression happening. That depends on the level you're going in with (in my case 0dBFS from the interface is at +13dBu) and how you calibrated the unit which also affects the output level somewhat. Obviously there's a lot more knob positions that can achieve unity gain as this is always a combination of both knobs AND the compression amount...

 

[Potato Cakes]

This is definitely not the compressor circuit where I would be concerned about precision signal control. As fripholm pointed out, the output is a moving target. I posted early on in this thread about the process I went through to match the channels for stereo operation and I found that even adjusting the release time affects things like THD. I spent a lot of time trying to figure out a procedure for matching before I settled on something that I thought measured and sounded correct. In the end I realized relying mostly on your ears and just a smidge on the meters goes a long way for this compressor. It's not a going to be particularly gentle with your audio tracks so don't get too caught up on really fine details. This will save you a bit of time and sanity and get you putting it to good use faster.

Here is a link to my post when I finished after going through a bit of struggle when building and calibrating it: https://groupdiy.com/index.php?topic=66219.msg874636#msg874636. Hopefully these will be some helpful things to consider when you start to wire it up and calibrate it.

For my stepped input and output controls, I just did a calculation based on the number of steps and whatever the level was for that position was what it was.

Also, this is my favorite compressor to use on electric guitars. And I have a bunch of compressors.

Thanks!

Paul

 

[octavez]

Thank you for clarifying it! Appreciate it!

 

[JMan]

Okay, I'm feeling dumb today.  I've been trying to wrap my head around the fine tuning procedures for a couple days (just in terms of which components to place, I haven't actually started the fine tuning process yet) and I want to make sure I am clear on this.

I have the revision v2.5 boards.

FT1 - all that is needed here is the 10k trimmer, right?  By my math, it looks like I should pre-trim it for approximately 4k ohms and adjust on test.  No fixed resistors for R7/8 in this case.

FT2A - You either place a 12k resistor in R22 and jump the middle and square pads of FT2A, OR you install a 10k resistor for R22 and a 20k trimmer for FT2A for adjustment. 

FT2B - A 2k trimmer (FT2B) is installed instead of R56.

My question on FT2A is, if the rough target resistance for R22 (or the total resistance between T5 and R25, as it were) is 12k, why does the second option use a 10k resistor and a 20k trimmer in series?  Wouldn't a 5k trimmer make more sense?  Or am I missing something here, because I'm feeling like I'm missing something...  I see that you mention in the guide that your right channel's final value for R22 was around 18k, so I guess the larger trimmer value is to compensate for wide swings like that?


And one final unrelated question: For C13, what are the pros/cons of using a tantalum instead of a regular electrolytic?  I have used tants in other builds when called for, but I always hear such conflicting information about whether they are good or bad.  Given the option here, I don't know what to choose or why.