U273b / U373a questions

[Murdock]

Hey folks,

since I got hold of a set of original output and interstange transformers for these I'm planing on building a clone.
I got the original manual which has some valuable informations. Such as the "official" mod for variable attack.
There is a thread here on Groupdiy where PRR posted a mod. But it's not correct.
To get variable attack on these you have to vary R34 (220 Ohm). Higher value or no resistor at all means quicker attack lower value means slower attack. When the resistor is omitted the attack is 30us. But they warn that such fast attack will rise distortion at low and mid frequencies. They advise a range from 100us to 10ms.

The manual also states, that no matter if the limiter is on or not the output is limited to +6dBu.
Can anyone tell me, where this limiting takes place?
I simulated the circuit in Spice and got it kind of working. The limiting is also active in the simulation. Or at least I see sinewaves getting cut off when going over +16dBu...
I tried a few things but I can't find the source for the limiting. My electronic knowledge is just too non existent to understand what's going on in this circuit...
I disconnected the sidechain and the limit was still there. So I assume it's in the makeup amp?
I can upload my Spice simulation if that helps.

PS: Just for info, U273b and U373a are basically the same except the second output transformer in the U273b for different output levels.
And there is an error in the floating schematics. There is no connection between C16 and R16.

 

[Murdock]

Nobody?

 

[AnthonyBisset]

@Murdock

Some years ago the studio owner at Widehive in Berkeley California (Gregory Howe) turned me onto the u73Bt. More recently, I heard on good authority that the u73Bt are just prototype u373a's, possibly with different transformers, although I'm wondering if the nuvistor tube claim comes because there were a number of different prototypes in the u73Bt run?

This thread's timing is great. Just last week I decided that the pain of cloning it would be worthwhile and started hunting for schematics. As impressive as Greg's u73Bt's are, the bass response could be optimized. Also, if the circuit is bringing some LPF behavior at 15Khz that could be removed since we're not in the FM days anymore.

If this project allowed the sidechain to be processed separately then a bit of delay from the DAW would enable this circuit to be a great look-ahead limiter. The u73Bt character reminded me of the middle ground between a comp and a soft clipper. I believe it's pretty ideal for mastering given a few optimizations.

 

[Murdock]

@Murdock

Some years ago the studio owner at Widehive in Berkeley California (Gregory Howe) turned me onto the u73Bt. More recently, I heard on good authority that the u73Bt are just prototype u373a's, possibly with different transformers, although I'm wondering if the nuvistor tube claim comes because there were a number of different prototypes in the u73Bt run?

This thread's timing is great. Just last week I decided that the pain of cloning it would be worthwhile and started hunting for schematics. As impressive as Greg's u73Bt's are, the bass response could be optimized. Also, if the circuit is bringing some LPF behavior at 15Khz that could be removed since we're not in the FM days anymore.

If this project allowed the sidechain to be processed separately then a bit of delay from the DAW would enable this circuit to be a great look-ahead limiter. The u73Bt character reminded me of the middle ground between a comp and a soft clipper. I believe it's pretty ideal for mastering given a few optimizations.

Hi Anthony,

The U73Bt found here (Reverb U73Bt) spots exactly the same circuit and transformers as the "normal" TAB U373a plus an internal power supply. So They are identical.
But it could be that earlier ones where different.

There are some spots in the circuit that tailor the frequency response.
For instance C9 (1uF) is chosen to have a basscut of -3dB at 20Hz. At least that shows my simulations. Making it larger increases bass response. These vary also in different revisions/modules.
1uF in the U373a. 3,3uF in the U373s and 22uF in the U373p (maybe that was 2.2uF and the dot was not printed right.
C15 (6,8uF) also affects bass response. In the U373s schematic I have it even says "Abgl." which means trim.

C11 and C6 are high cuts. Making them lower or removing them increases high frequency response.

Are you trained in electronics? Do you have an idea where the liming is taking place?

 

[AnthonyBisset]

Hi Murdock,

Thanks for confirming the u373a and u73Bt are the same. The magic I heard with the u73Bt is worth the effort in my opinion and it sounds straight forward to improve the frequency response.

I don't play at the level most do on this forum (with regards to EE and circuit design) and I'm just now working at understanding how this circuit functions. So, I'd love to spend some time with your spice model and reduce my ignorance.

From my sleuthing I've gathered this design uses transistors as diode bridge elements. I don't know how it would improve the circuit but bipolar transistors can produce "very low leakage diodes" --(Thanks stack exchange).

@Heikki on this forum made an interesting diode bridge compressor, which in my conversation with @scott2000 suggested it was loosely based on the u273/373a (and of course the Neve 2254). Thread is here:

https://groupdiy.com/threads/another-diode-bridge-compressor.79706/

An interesting aspect of these designs is the noise floor can be reduced by -6dB per each doubling of diode elements. I'm not sure that's practical here (or even desired) as the primary source of magic i heard in my friend's u73Bt was a euphonic clarifying of the midrange after percussive transients caused significant gain reduction. As the compression eased up there was a kind of sparkle or fractal harmonic distribution that would bloom out the midrange before the next transient. In this case in between kick and tom elements.

Perhaps an ideal design would allow one to tune how much of this friendly/euphonic distortion presents in between heavy compressive moments. I could imagine varying the total diode stage count based on an envelope follower tracking the content so that more complex passages would present less distortion while sparser midrange focused passages would have more harmonic bloom. Since the effect really only occurs after transient events the delay of an envelope follower might be fine, or it could be another use case for a digital delay. Wild ideas aside, I'm excited about this circuit in its original form. Thanks for digging into the design.

 

[AnthonyBisset]

Yesterday I purchased some 373a and 373p modules in a nice chassis. If any questions about components come up I can open a module.

 

[13engrsapper]

There is an unracked set of U373's here. I will check which version. We sold the U273's we had - they sounded great, we still have a set of U73's.

 

[Murdock]

Yesterday I purchased some 373a and 373p modules in a nice chassis. If any questions about components come up I can open a module.

There is an unracked set of U373's here. I will check which version. We sold the U273's we had - they sounded great, we still have a set of U73's.

Thanks for the info and offer. Will probably come in handy to have originals for measurements.

In the meantime I found out, that the supply voltage is one part of the "culprit". If I increase it to 35V without changing anything else, I get no clipping.
But as 24v should suffice to get clean signal up to 22dBu it has probably to do with the biasing of the transistors in the amp stage.
As I really don't know anything about biasing transistors I have to read some stuff about it and see if I can come up with something.
And then take it over to the real world and see how it goes.