Reverse Engineering an AN/GSA33 Navy Compressor

[vinyvamos]

In the past days I have been working my way through this piece of old US Navy equipment from about 1961, which is essentially five compressor modules in a rack with trafo balanced ins and outs. The frequency range is quoted at 200-5KHz on the output trafos, but both input and output iron sweep perfectly from 20-20K. This is not the first time I have tested MIL-SPEC iron to be way better than stated. The F.R. when measured was more like 140-10KHz (-3dB), so the limited range must be in the circuitry. I am now trying to increase some coupling cap values to at least lower that crap LF shelf... Anyway, back to that later...

What is very unusual about this design is that it seems to use a network of varistors for gain reduction in front of the first amp stage. Can this really be? Am I missing something essential here? They are all reading in the Mega ohms range when tested in circuit. I have read one discussion on here about the uselessness of them for this purpose...

https://groupdiy.com/index.php?topic=26880.0

The designators on the board are RV201, RV202, etc and you can see them in the attached image. The dark metallic grey discs in two sets of three on the board. I will also post an image of the entire schematic if anyone is interested. It is all PNP germanium transistors of just one type; 2N461. The main supply is a regulated -30V while there is a low current zener reg. +30V being provided to the sidechain section.

 

[vinyvamos]

Front of the rack. These have been discussed before over at the Gearslutz forum, with little or no info available on them, and nobody knowing if they are worth bothering with for studio use

 

[emrr]

Nice drawing.  Have seen those sold for decades, never any comment about usefulness.  Nothing immediately to add other than pointing out the reduced bandwidth may be to prevent or hide compression artifacts.  Several other Govt types I've tried to modify would go more full range, but got ugly in doing so. 

 

[vinyvamos]

Thanks Doug . Here is the latest version as I had missed two resistors in the sidechain! I knew there was something a-miss. I deleted the previous Schematic post to avoid confusion.

I have started to work through one module with a test probe, and it seems that the LF is being lost in the final output amp board. Upon increasing C215 to 2uF and C216 to at least 100uF I now have the -3dB point at 40Hz... Still more digging to do, and maybe the HF can be helped by some boosting somewhere. If I don't load the output with 600 ohms and leave it open the F.R. is generally better and the HF extends to 15KHz. The modules sound pretty decent now actually. The release is pretty quick and the attack is quite smooth, and it sounds really good, yet a tad dirty, on individual instruments. On a full mix though it tends to get messy with a lot of breathing/pumping.

 

[mjrippe]

Very excited to see this, I have two of these modules on my shelf!  Do you have a simple pinout for the Amphenol connector?

 

[vinyvamos]

Very excited to see this, I have two of these modules on my shelf!  Do you have a simple pinout for the Amphenol connector?

Yes I think quite a few people have these modules, or even a complete rack of them gathering dust, or just asking to be modified .

Here is the AMP connector pinout:

1 - Input +
2 - Input centre tap
3 - Input -
6 - Output +
7 - Output centre tap
8 - Output -
11 - GND/0V
12 - Reg. +28V (low current, just provides biasing for sidechain)
16 - Reg. -30V (probably 100mA will do here, I haven't measured any current draws yet)

If you increase the cap values of C215 and C216 on the output board then you will get much better LF response, and don't bother loading the output when feeding into modern hi-z balanced inputs .

There is also questionable grounding topology in these modules. 0V is connected to the rack chassis at the power supply, and then it is also connected to the module chassis at two points within the module. Not textbook I would say, and I did see small improvements in one module when I changed it to a single point GND within the module.

 

[mjrippe]

Much appreciated!  I had it all sorted except for that bias voltage  :  For some reason the documentation (NAVSHIPS 93433) has never shown up online.

One of mine is made by Stromberg-Carlson and the other is by Melcor (yes, THAT Melcor).  Anyone else make them that you know of?

 

[TubeMonkey]

Hey guys,

I have two Melcor versions which I am about to wire up. I've replaced the McMurdo connectors where someone didn't have the female plugs so they decided to direct solder to the male connectors.

It seems all of the attachments in the entirety of Group Diy are returning a 404 error when I try to open them. Is anyone else experiencing this? Would somebody be kind enough DM me the schematic that was posted up top?

Thanks!

Ethan Barrette