3k9 vs 5k1

[osterchrisi]

So, now for the last time and ultimate knowledge: Why why why are the THATs on the GSSL boards connected to GND via 3k9 resistors? The THAT datasheet tells us to use 5k1 here, but the resistor hasn't been changed in any revision. Why is the 3k9 suggested? Isn't it moving the operation point of the VCA?

Thanks for your answers, this gives me sleeplees nights...  :-\

 

[pedroplanet]

For That 2180, I've been using 5k1 on the DBX202 emulation circuit, but not on the sidechain (i'm still in doubt about it)

The schematics/parts layout are outdated and probably won't be updated... I think no one came with a FINAL solution about it.

some links I found
http://diy.fischerworks.com/gssl_vca.shtml
http://homepage.mac.com/marten.thielges/gssl/calibration.html

I need help with turbo board and 2180-resistors change (yes the turbo board has 3k9 also)

 

[osterchrisi]

Oh yeah thanks a lot for your efforts. I know all these pages and already built the GSSL and it sounds great.
I simply wonder all night long, why the heck the 3k9 came into play anyway when the datasheet says something different. I mean, there has to be a REASON for that, I guess Gustav didn't pick random resistor values 

 

[pedroplanet]

Certainly, but I think he made it for the original DBX202 and 2150's... (by the way, thank you Jakob!)

the new 2180 are a bit confusing here...

My Gssl is working good, but not perfect yet,
using the turbo board and trying to do the same as on the video, my compressor tends to compress a bit more on the left side.

I'm still in doubt about the 100k (ratio ajustment) my compressor seems to be a bit smother than I think it shoud...

How did you make yours? what resistor changes did you do?

 

[osterchrisi]

I more or less changed the resistors suggested at Matt's Page. The Ratio Adjustment resistor I changed to 120k, and also the 3k9 from the signal VCAs but not the sidechain nor the Turbo Board VCA. I used 2180s. Didn't solder all the trimpot related resistors. I will try to change the two remaining 3k9s also to 5k1 but I'm building this for my brother, so I want him to hear the changes and then agree about keeping it or dismissing the change.

 

[pedroplanet]

on Ratio I used 127k as on the last parts Layout.

Cut the pin 4 on all 2180's

used 5k1 and removed the 68R on dbx202 emulation circuit.

 

[PRR]

> Isn't it moving the operation point of the VCA?

Not 99% clear I know which of the many variants you mean.

Yes, this resistor sets cell current.

The bias current must be greater than the signal current.

Plus some margin so it don't get close to clipping.

Signal current is set by resistor into the VCA and NFB resistor out of the VCA.

Because there are multiple paths inside the VCA, "bias" current may have to be set 2X or more times the actual cell current... I don't care to untangle that topology tonight.

If I read it right, you are advised to use 3K9 bias and 15K NFB around the following amp.

THAT Corp sheets suggest 5K1 and 20K.

15/3.9= 3.85
20/5.1= 3.92

The peak/bias ratio is "nearly 4" in either case. The 2% difference for the two sets of values is utterly meaningless. "4" isn't an especially magic number.

The 3K9/15K values will eat a hair more power and give a few dB less hiss, but THD will rise a bit. This is documented in the datasheet page 6.

This was NOT worth sleepless nights. Good night.

 

[pedroplanet]

so using a 5k1 and not changing the 15k out would make the things even worse right?
(as I did)

thank you very much for the explanation 

 

[angelo2979]

Sorry to resurrect this old thread, I'm about to finish yet another Gssl, so I was collecting all the info to be sure not to forget anything.

but I found here is some interesting information... if just anyone could confirm the statement above, about using 5k1 with a 15K resistor to pin 1, making things worst or not, instead of 5k1 + 20k as suggested by THAT corp ...

what really confuses me is why "newer gssls" have new board layouts, with 5k1 resistors in place of 3k9, but still that same old 15k to pin 1...why not change to 20k resistors? What's better?

I guess this is a bias change that could affect the tone, more or less noticeably (?), maybe a minor change, but interesting to know and learn


please let me know,
thank you.

 

[abbey road d enfer]

but I found here is some interesting information... if just anyone could confirm the statement above, about using 5k1 with a 15K resistor to pin 1, making things worst or not, instead of 5k1 + 20k as suggested by THAT corp ...

Not worst, different. Choosing the operating point of a VCA cell is always a trade-off between noise and THD.
Most people are more sensitive to noise than distortion, so I can understand the choice, but at the end of the day, the difference is about 20%, so 2dB less noise for 0.006% THD instead of 0.005. Still it is much less than the distortion induced by side-chain modulation.

 

[Harpo]

what really confuses me is why "newer gssls" have new board layouts, with 5k1 resistors in place of 3k9, but still that same old 15k to pin 1...why not change to 20k resistors? What's better?

The 5K1 and 3K9 have nothing to do with the VCAs pin1 connection. Each of these resistors connects between the VCAs pin5 and its negative supply rail (not GND, as wrong said in the 1st.post) in order to set the bias current. The THAT218x in the DBX202C substitution circuit is supplied by +/-15VDC, using the 5K1. The THAT218x in the sidechain is supplied by +/-12VDC, using the 3K9 instead, setting this type of VCA for same 2.4mA bias current. Maybe read the VCAs datasheet for an in detail explanation.

 

[angelo2979]

The 5K1 and 3K9 have nothing to do with the VCAs pin1 connection.

Of course, I was talking about the 15k resistor, which does connect to pin 1 of all the VCAs in the board. If I'm not wrong, the Datasheet suggests this to be a 20k resistor, where between pin 5 and the negative rail is suggested a 5k1, as you say.

The THAT218x in the sidechain is supplied by +/-12VDC, using the 3K9 instead

This confuses me further. Newer board layouts have 5k1 all around. Can't see that 3k9 exception on the sidechain, as you can see it here https://pcbgrinder.com/download/GSSL_11.pdf 
This is the latest revision people are building with 218x. You can see all 5k1 resistors to pin 5, and 15k resistors to pin 1 for all VCAs. Similarly turbo boards ship with the 3k9 + 15k combo, which got to be changed accordingly.

If PRR's old post is correct, pairing 5k1 with 15k instead of 20k may be less then an improvement, meaning new layouts are wrong...

THAT Corp sheets suggest 5K1 and 20K.

15/3.9= 3.85
20/5.1= 3.92

The peak/bias ratio is "nearly 4" in either case. The 2% difference for the two sets of values is utterly meaningless. "4" isn't an especially magic number.

The 3K9/15K values will eat a hair more power and give a few dB less hiss, but THD will rise a bit. This is documented in the datasheet page 6.

Hope that makes sense, let me know what you think! thank you!

 

[abbey road d enfer]

If PRR's old post is correct, pairing 5k1 with 15k instead of 20k may be less then an improvement, meaning new layouts are wrong...

It depends very much on the actual signal level. If the level scaling before the VCA reduces the amplitude, the 4:1 ratio between cell current and peak audio current may be altered.
I've often used 15k-4.7k when noise was a major concern.