All-discrete GSSL signal path upgrade

[SSLtech]

A couple more new DIY guy questions to pester with which to pester you...

No worries!

In the photos posted on this thread there are no resistors or radial capacitors in the output section of the GSSL. I didn't see that mentioned in the manual. Do I also need to leave out these resistors and capacitors? I also noticed pins 1 and 2 are not connected on the DIP8 socket at the output position on the main board. Are these not being used? Will it affect anything if they are connected to the Cavendish with the appropriate pins swapped?

No, they don't do anything, there's no connection to the Discrete op-amp. In the original transformerless output, there are TWO op-amps used to generate the opposing polarities for the output signal. The Cavendish uses only one op-amp, and balances it with a transformer, so there's no real reason to worry about the other pins.

In fact in mine, I connected only five leads between the two, but if you do it that way, you HAVE to remember which way round the jumper cable goes... otherwise if you reverse it, there's no connection where there needs to be one, and a connection where there doesn't need to be one... so I thought I'd just advise people to fully-transpose, which means that the lead can be used either-way-round.

Hope that makes sense!

Additionally, the manual mentions removing the 7.5K resistors on the main board and use the 5K pots and trimmer on the Cavendish. I cannot seem to locate any resistors with that value. Is this not already done with cables jumping the DIP8 sockets on the different boards?

Depending on the revision level of the board you have, the resistors MAY be marked as 15kΩ resistors. The output gain ratio is calculated by dividing the value of the resistor going INTO the VCA (usually 27k) into the value if the resistor in the inverting feedback loop on the folllowing op-amp, all multiplied by two.

So, 30k (rounding up) divided into 15k = 0.5, multiplied by two = unity (1).

So, If you have 27k resistors going in to the VCA, you should really use 13.5kΩ in the feedback loop, so either a fixed 15k (as in the original version) and the error will be less than a dB, but usually not quite perfect unity gain... or use a 10k resistor and a 10k trim pot. the middle of the trim range will give you 15k... pretty close, and a little tweak should allow you to dial it absolutely spot on.

If in doubt as to which resistors not to fit on the main board, it's the one on each channel which provides the negative feedback from the first op-amp output back to the input node (which is also hard- connected to the output of the VCA).

Let me know if that confuses more than it clarifies!

Keef

 

[Potato Cakes]

Thanks for your response! This also answers my question regarding the polarity and how to wire the outputs to XLRs.

My GSSL board actually has the 15K resistors. Just to clarify, should I remove them since I have installed a 5K resistor with the 5K trimmer as the manual recommends?

Pending one last Mouser order, I hope to have this done and pictures posted soon. Thank you again!

 

[SSLtech]

Well, if your input resistor (feeding the input of the VCA) is 27k, then you should indeed remove the 15k resistor, and replace it with 10k resistors and 10k trimmers, instead of 5k resistors and 5k trimmers... double check, but yes.

The only issue would be with output gain... you'd be low (or high) and lacking the 'adjustment range' to achieve unity gain.

If you've already ordered the 5k resistors and trimmers, you could always tag a 7k5 resistor in series with the 5k, and still get to unity, with a finer adjustment range...

 

[Potato Cakes]

Thanks! I'll give it a go and report back.

 

[Yannic1990]

Hey Guys,

after finally finishing the turbo mod i want to build the cavendish board.
BTW the turbo sounds awesome 

I bought the EA 2503 from Classic API but the cables aren't the same order as
in the manual.

What is the right way to connect it to the pcb?

Manual      I      My EA2503
---------------------------------
Brown                  Brown
Red                        Orange
Orange                Green
Yellow                  Violet
Green                  Red
Blue                      Yellow
Violet                    Blue
Grey                      Grey


I hope someone can help.

Thanks

Yannic

 

[martthie_08]

@Yannic the order of cables coming out of the transformer is completely irrelevant. Just make sure to hook it up the right way: brown/red is primary, orange/yellow secondary #1, green/blue secondary #2, violet/grey secondary #3

@SSLTech: very nice project, I have been having an itch for a discrete SSL comp for quite some time and just noticed your wonderful project, congratulations!

It would be interesting  to place your discrete stages on the in / out of the "standard" IC based unit and compare that to your modified discrete signal path.

side note: IIRC the Vertigo Sound compressor went an interesting route in their own way, making a reproduction discrete VCA and using a DRV134 for balancing out and a Jensen transformer on the input.

 

[finbase]

Greetings!

I'm working on the Cavendish mod right now. But since I'm not a native English speaker, I just want ensure that I've understood everything right.

If I'm using 10k resistors and 10k trim pots on the Cavendish board, I'd have to take out the 15k resistors on the main board's output section, right? So please see the attached picture, I assume that depicted resistors are the ones that has to be removed?

EDIT: Oh, and are 100uf caps OK as C9 and C10 on the Cavendish board (I suppose so, if the caps are meant to be as large as possible)? I'd happen to have some Elna Silmic IIs laying around of that value...

Thanks!
finbase

 

[SSLtech]

Yes, those are the 15k resistors to remove... and yes, API (for example!) used several different cap sizes for the DC block to the transformer. 470µF was one, bypassed by 47Ω, (which acted as a 'fuse' if the op-amp went bad!) and all sorts of others.

Some brave souls may want to short past the DC blocking capacitor... but not me. I'm very risk-averse!

As for the transformer lead color sequence... As correctly noted earlier, the COLORS are what matters. Ignore the sequence that they physically exit the transformer.

Keef

 

[martthie_08]

API (for example!) used several different cap sizes for the DC block to the transformer. 470µF was one, bypassed by 47Ω, (which acted as a 'fuse' if the op-amp went bad!) and all sorts of others.

what's a useful Wattage rating for the 47R on the output? Guessing 1W? any higher and the fuse function will be lost?
although slightly OT but I DID have a fried 2503 on a vintage 550 EQ that didn't have this resistor...

 

[finbase]

Hooray hooray!

I just finished my Cavendish moded GSSL! O boy I'm glad! It  worked almost straight away (needed to double-check my grounding connections, but it was an easy task)!

Works great and sounds GORGEOUS! I'll post some pics soon! 

I'll also do some AB-testing with my "regular" GSSL as soon as I get it back. I lent it to a friend of mine for a test drive. If anyone is interested, I'll record some comparison samples.

Thanks to everyone who has helped along the way! And special thanks to Expat Audio for inspiring project!

 

[finbase]

Here's the pic I promised! (I'd have more, but I can't attache them on the text field...)

Cinemag CMOQ-2S trannies, Sound Skulptor SK25 op amps and Elna Silmic II caps in the signal path. Brilliant!

I finally got my "regular" GSSL back, and I've done some A/B testing with the Cavendish one... The differences are subtle, but still  perceptible. It's always a pain to describe sonic things in words, but I'll give it a go. Please bare in mind: this is all very subjective!   

Cavendish feels more even across the spectrum. It seems to create a wider and more open feel, it "breathes" a tad more. Low end is a bit "bloomier", and sounds a bit more natural to my ears. It does feel bigger and deeper, as SSLtech said.

But I like especially the high end: the top end sound much more interesting, Cavendish makes the high end really sparkling. The regular GSSL feels a bit more "plastic" as far as it's low- and high-end is concerned, but "plastic" in a good sense... Like Bubble-Gum-Pop-Music sense! 

Nevertheless, the most prominent difference is in the middle frequencies: the regular GSSL is more forward-sounding, it has a notable "mid-bite". Cavendish is more even, and as I said, it "breathes" more, giving a bit more open feel. But I do like the regular GSSL's "attitude" also!

How much do they differ? I guess that the importance of those subtle differences is highly subjective... But at least to me they're significant enough that I want to keep them both!

 

[martthie_08]

thanks for your thoughts about the discrete in/out stages!

do both units have identical parts throughout (minus cavendish mod)?

have you thought about going over to discrete VCA's to make the signal path fully discrete?

 

[finbase]

thanks for your thoughts about the discrete in/out stages!

do both units have identical parts throughout (minus cavendish mod)?

have you thought about going over to discrete VCA's to make the signal path fully discrete?

Almost exactly the same parts, but those Elna Silmic caps are only in my Cavendish version. The elcos in the regular GSSL might be Nichicons, although I'm not 100% sure...

That discrete VCA sounds interesting option. Although I'm a bit sceptical whether one is going to achieve much improvement or not...  Do you happen to know some particulat discrete VCA project, that would in your opinion fit with this Cavendish mod?

Best regards,
finbase

 

[martthie_08]

finbase, thanks for the info!

fyi I have ELNA Starget in my GSSL and am really quite happy with those caps, they did make a noticeable difference to the off-the shelf noname stuff.

regarding the discrete VCA's - I am about to do some experimentation in that field and have both gold and black can DBX 202 coming to my place next week. I had a chance to beta-test an early prototype of the Vertigo Sound VSC-2 and the guys in Munich put quite  some effort into manufacturing a discrete clone of the gold cans, which they thought was worth it. I personally have not looked into DIY-ing one, but opted to score some old originals...

 

[finbase]

finbase, thanks for the info!

fyi I have ELNA Starget in my GSSL and am really quite happy with those caps, they did make a noticeable difference to the off-the shelf noname stuff.

regarding the discrete VCA's - I am about to do some experimentation in that field and have both gold and black can DBX 202 coming to my place next week. I had a chance to beta-test an early prototype of the Vertigo Sound VSC-2 and the guys in Munich put quite  some effort into manufacturing a discrete clone of the gold cans, which they thought was worth it. I personally have not looked into DIY-ing one, but opted to score some old originals...

Well that VCA experimentation sounds interesting. Would it be too much to ask if you could report about the results via an PM? I'd be very interested!

I just thought that maybe this VCA discussion is getting this Cavendish dedicated thread a bit off-the-topic...

Best regards,
finbase

 

[martthie_08]

I just thought that maybe this VCA discussion is getting this Cavendish dedicated thread a bit off-the-topic...

I completely agree, it is so easy to get sidetracked, although this thread title kind of tempted me to do so  8)

sure, of course I will let you know about my findings, it might take some time though, I am just going into a 2-week tracking session out of town...

 

[finbase]

I just thought that maybe this VCA discussion is getting this Cavendish dedicated thread a bit off-the-topic...

I completely agree, it is so easy to get sidetracked, although this thread title kind of tempted me to do so  8)

sure, of course I will let you know about my findings, it might take some time though, I am just going into a 2-week tracking session out of town...

Well, as far as ANY inspiring gear is concerned, it's both easy and fun to get sidetracked...  ;D I think that every single person that has been drawn into the audio gear DIYing is already extremely sidetracked. We're not normal, you know! Just ask from your wife! 

No worries,  take your time! I'm not in a hurry! Thank in advance!

Regards,
finbase

 

[Whoops]

Many compliments to the "Designers"

any audio sample "comparison" between  "standard" Gssl and the  "Apized"  ?

 

[Potato Cakes]

So it seems that my Cavendish is causing a buzz with the ground connected to the star ground and a 60Hz hum when it's disconnected. I messed around with different grounding combinations, but I wind up with the same results. When the it is out of the signal path there is no noise. I placed the Cavendish board right next to GSSL PCB with the transformers facing the rear towards the XLR input connectors. I noticed that other people have it turned with the DIP8 connections facing the GSSL PCB. Would the orientation of the Cavendish to the GSSL board cause this sort of issue?  It passes audio just fine, and the noise compared to the signal wouldn't be heard in a mix for most things. But if being used for a stereo mix, it would be noticeable during quite passages if listening at a decent volumes.

Thanks!

Paul

 

[PeterZweiMeter]

hey guys,

will be the gssl cavendish in stock again?

thx
peter