Choosing transformers for RM58 compressor a la Record Plant

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

soapfoot

Well-known member
Joined
Dec 27, 2010
Messages
1,062
Location
LA + Brooklyn
I'm putting together a plan for a Roger Mayer compressor, and I want to put transformers on the inputs and outputs like was done to the one at the old Record Plant.

That one went up for sale recently, and I was able to grab this pic--it shows that the output transformers were UTC A-21s.

However, the input transformers (also UTC A series) are rotated such that you can't see the part number.

I'm trying to deduce what would be the best here-- more A-21s? Or perhaps a 500:50k like an A10 (or clone)?

I'll attach some photos of the Record Plant unit and relevant RM58 documentation

 

Attachments

  • Screen Shot 2021-01-06 at 1.17.49 PM.png
    Screen Shot 2021-01-06 at 1.17.49 PM.png
    331.1 KB · Views: 77
And finally, a shot of the original with the UTC iron added on:

 

Attachments

  • Screen Shot 2021-01-06 at 1.17.03 PM.jpg
    Screen Shot 2021-01-06 at 1.17.03 PM.jpg
    483 KB · Views: 54
I guess if it needs gain on the input, that would be a clue.  With the 100K pot I'd tend to agree 10K:10K seems to make sense, but it could be many things.  600:600 doesn't make a lot of sense with the 100K, but you never know. 
 
Rob Flinn said:
By the way did you see that this has already been cloned ? If not take a look in the dynamics section.

I did, thanks! That was the starting point for my build. I'm making nice Gerber files in CAD for PCB art, I'll share them once I get it working, tested and verified.

UTC didn't appear to make a 10k:10k in the A style case, so it seems as though the pictured unit from Record Plant must've used something else.

The A16 is 15k:60k... Maybe that's a candidate?

The A24 is 15k:600. The highest impedance 1:1 seems to be 600:600
.
 
Another thought-- the Record Plant unit had its potentiometers replaced, as well.

I wonder if that 100k pot was replaced with something else? Any ideas for a UTC transformer and pot combo that would work well?
 
A-24 backwards crossed my mind.  RCA used it that way as an API type preamp input.
 
EmRR said:
A-24 backwards crossed my mind.  RCA used it that way as an API type preamp input.

Ah, now that's a thought. 600:15k then, right? Is my math correct that that would result in a gain of nominally 28dB?

Speaking of the pots, I'm frustratingly close to a clue here from these photos (attached)

The input pots are actually visible in some of the photos I kept. Zooming in, you can piece together that the rear seems to say "721304 - 003"

This is definitely not an EIA code because it doesn't match A-B's EIA number. Apparently the model numbers stamped on the rear of the Type J pots changed a few times, and none of the datasheets I can find have helped me to decode this varietal. It's tempting to assume that the "304" means "300k" but that's far from conclusive, and besides I don't think that was a standard value for the Type J (35k was).

Additionally, you can almost see how the mystery transformers are connected. I'll scrutinize this a bit more later and see what I can come up with.
 

Attachments

  • RM_CLUES.jpg
    RM_CLUES.jpg
    939.1 KB · Views: 38
Ok, that makes me think it's more A-21's and they're rotated in an attempt to reduce crosstalk. 

A-24's are 15K:500 turns 5.477 gain -14.77dB  I think, in theory, somewhere it's written as 13dB, since there are losses to account for.
 
EmRR said:
Ok, that makes me think it's more A-21's and they're rotated in an attempt to reduce crosstalk. 

A-24's are 15K:500 turns 5.477 gain -14.77dB  I think, in theory, somewhere it's written as 13dB, since there are losses to account for.

Yeah, I'm starting to lean toward the conclusions that it may be a quartet of A-21s myself
 
soapfoot said:
Ah, now that's a thought. 600:15k then, right? Is my math correct that that would result in a gain of nominally 28dB?
You don't want any gain here. The circuit is designed to operate at about -20dBu (remember it was originally a guitar pedal!). A10:10k xfmr is perfect, since the pot+input network provide the necessary attenuation.

The input pots are actually visible in some of the photos I kept. Zooming in, you can piece together that the rear seems to say "721304 - 003"
that sounds to me like it was manufactured 13th week of 1975.

  It's tempting to assume that the "304" means "300k" but that's far from conclusive, and besides I don't think that was a standard value for the Type J (35k was).
Don't sweat over this. The load is about 110kohms, so any pot between 20k and 200k would work satisfactorily.

Additionally, you can almost see how the mystery transformers are connected. I'll scrutinize this a bit more later and see what I can come up with.
I'm rather dubious about an output xfmr. As it is, the output stage is designed for light loads, like 10kohms. Are you sure there wasn't a separate buffer for driving a xfmr? Note that a 10k:10k would work, as long as the cable run is not excessive.
 
abbey road d enfer said:
I'm rather dubious about an output xfmr. As it is, the output stage is designed for light loads, like 10kohms. Are you sure there wasn't a separate buffer for driving a xfmr? Note that a 10k:10k would work, as long as the cable run is not excessive.

Thanks, I'd be interested to hear more. Seems like the for sale listing that I grabbed the pics from is still available, and I'll post it here. No evidence of added booster amp

This was actually an RM57, but the two are very similar.

I'd welcome any additional insights!

https://reverb.com/item/11866606-crazy-rare-roger-mayer-rm-57-stereo-compressor-from-the-record-plant-in-nyc-modded-bra
 
Confirming that those are A-21s. Pulled one out and they have a pin arrangement unlike any other A series. Seven pins total, 3 grouped together, two opposite those and 2 on the outside center near the edges. See below.

xsxepjtl.jpg
 
rackmonkey said:
Confirming that those are A-21s. Pulled one out and they have a pin arrangement unlike any other A series. Seven pins total, 3 grouped together, two opposite those and 2 on the outside center near the edges. See below.

xsxepjtl.jpg

Thanks so much for that info!

What's interesting is that it seems, according to many here who are better-educated than me, that these would not necessarily be the optimal choice for both input and output.

I'm eager to understand any hypothetical negative unintended consequences of using four A-21s in this application. I've been reading Whitlock papers all night but I still lack the sufficient background to really speculate on what might be in store.
 
soapfoot said:
I'm eager to understand any hypothetical negative unintended consequences of using four A-21s in this application.
As an input xfmr, the A21 does not have much headroom, which implies that the send level must be constraint, which should be the role of a compressor... ???
As an output xfmr, in this particular case, headroom is not an issue, since the output stage cannot deliver the voltage, but it also cannot deliver the current needed to drive  the xfmr at low frequencies.
So in that case, it's not an issue, but it would be if the xfmrs were used with a unit that has headroom.
The consequence is that this unit must be run at reduced level; this is the logic consequence of using a guitar FX circuit without the necessary input/output buffers.
 
abbey road d enfer said:
As an input xfmr, the A21 does not have much headroom, which implies that the send level must be constraint, which should be the role of a compressor... ???
As an output xfmr, in this particular case, headroom is not an issue, since the output stage cannot deliver the voltage, but it also cannot deliver the current needed to drive  the xfmr at low frequencies.
So in that case, it's not an issue, but it would be if the xfmrs were used with a unit that has headroom.
The consequence is that this unit must be run at reduced level; this is the logic consequence of using a guitar FX circuit without the necessary input/output buffers.

I think I'm starting to understand.

So as an input, it would core-saturate and contribute distortion through that mechanism? And as an output it may work okay, or may impact frequency response?

One of the barriers to my understanding at this point is that, while I know there's a difference between a 10k:10k and a 600:600 transformer (despite both being a 1:1 ratio), I'm not entirely sure what that difference is, which makes it hard to understand why one would have more headroom than the other.
 
The stated level on a transformer is at the lowest frequency specified, and is vastly higher as freq goes up.  Much higher inductance is required as impedance goes up, a 10k:10k apples to apples against a 600:600 has much higher inductance.  I’m failing to see how this 600:600 has insufficient headroom here; consider expected working levels and drive sources.  It may have treble ringing from misloading, but being a 600:600 UTC, I doubt it.
 
EmRR said:
The stated level on a transformer is at the lowest frequency specified, and is vastly higher as freq goes up.  Much higher inductance is required as impedance goes up, a 10k:10k apples to apples against a 600:600 has much higher inductance.  I’m failing to see how this 600:600 has insufficient headroom here; consider expected working levels and drive sources.  It may have treble ringing from misloading, but being a 600:600 UTC, I doubt it.

If there was ringing from mis-loading, is this something someone might hypothetically bodge on a Zobel network to tame?
 
Back
Top