Rohde und Schwarz Begrenzungsverstärker ABR / U23

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

Falk

Well-known member
Joined
Jun 30, 2013
Messages
156
Good evening gentlemen,
I am planning a diy take on this one and gathered many information. Still missing are the information on the transformers. Especially ratios and impedances are of my interest. Also some information on the three inductors used in the tube stages and psu would be great.

I am located in Berlin, Germany. If anyone around has a unit at hand and would allow for measurements I d be extremely thankful.

Kind regards,
Falk // blankroomaudio.com

 
Falk said:
Good evening Gentleman,
I am planning a diy take on this one and gathered many information.   

Hi Falk,
Ooh, ooh, that's a nice unit.  I don't have any useful info at hand, just wanted to give a thumbs up  :)

Do you have a basic schema you're working from that you could post?
Transformer ratios and inductor values could probably be approximated as place holders for now at least.
 
Winston O'Boogie said:
Hi Falk,
Ooh, ooh, that's a nice unit.  I don't have any useful info at hand, just wanted to give a thumbs up  :)

Do you have a basic schema you're working from that you could post?
Transformer ratios and inductor values could probably be approximated as place holders for now at least.


See attached
 

Attachments

  • rohde & Schwartz limiter.pdf
    52.1 KB · Views: 339
Cool, thanks Rob.
Shooting entirely from the hip, given I don't know any of the passive component values or what valves are in this:
I would guess the interstage and  input transformers are 1:1
Output transformer  4:1 or 8:1, side-chain output 4;1
Anode load choke won't need to be as big as say a V72 choke (700+ H) since the pentode here is strapped for triode so, 200H?
With the values of the resistors in hand, a  better approximation could probably be made based on the throughput gain

Isn't there a dutch company that was building a new/improved version of this at one point?  Might be worth checking to see.

Edit - revised opinion on ratios and also found the company making new units here:
http://vacuvox.com/vacuvox-u23-tube-compressor-limiter/

 
Winston O'Boogie said:
Isn't there a dutch company that was building a new/improved version of this at one point?  Might be worth checking to see.

Edit - revised opinion on ratios and also found the company making new units here:
http://vacuvox.com/vacuvox-u23-tube-compressor-limiter/

Maybe it doesn't sound so much like a U23 ?

"The circuit was also modified resulting in a more stable operation over a much broader range. The lowest possible distortion was not a design goal during development and reducing the gain results in some higher second and third harmonics, which give the effect of making the sound appear louder and more euphoric."
 
back to point, here's some of what I have on the topic..

this is an excerpt from the legendary Braunbuch, whole article is over our upload limit

/Jakob E.
 

Attachments

  • Begrenzungsverstärker U23.pdf
    746.9 KB · Views: 223
I admit I didn't read the blurb fellas.  I was mainly thinking that Falk might manage to extract info about transformers from them.

Is there a scanned copy of the original schematic anywhere?  I looked on the Kubarth site but that particular schem  link is broken.  Also can't seem to find the docs that were once available on the IRT site.

Edit: I see Jakob posted while I was typing.  Thanks  :)

OK, so I see there's a fairly stiff pad on the front end and the values point towards the input transformer primary being for a low impedance source so, maybe that transformer isn't a 1:1 but more along the lines of a mic to push-pull grids ratio.
 
Good Morning,

thank you everybody for the valuable information. I am working on a spice model at the moment. At this point I only achieved a model of the three EF86 / EF804s stage and the PSU. With the values of the original schematic this stage has a gain of 1 with a 10k : 600r OT. Using 200H inductors for the plate choke and anode works well. Even though a 100Hz sine already looks slightly distorted. 1kHz looks clean.  I will work out this model as far as possible.

With respect, I assume the Vacuvox U23 design mainly allows for settings on the front panel that were originally on the inside and not needed at hand in broadcast use.

My plan is to use Edcore TXs and Hammond chokes. I will build a first version with turret boards and point to point wiring. As soon as I get it running I will layout a pcb.

Would you rather skip the input pad and use a 1:1 TX on the input? Could you explain the advantage of having the four secondaries of the interstage TX wired in series rather than having a single winding?

Best regards,
Falk

Edit: Got the OP stage fixed in spice now. Works like a charm. Plotted is input vs output at 100Hz. Frequency response is okay but could need fine tuning at some point.

 

Attachments

  • OP_U23.png
    OP_U23.png
    331.5 KB · Views: 127
I am working on a spice model at the moment. At this point I only achieved a model of the three EF86 / EF804s stage and the PSU. With the values of the original schematic this stage has a gain of 1 with a 10k : 600r OT.

OK nice.  So 4:1 works from a gain standpoint and, with the follower beforehand, probably no need for anything like an 8:1.  Most of the designed for NWDR units I've played with seem to have really quite low output impedance, in the region of 30R if that helps.

Using 200H inductors for the plate choke and anode works well. Even though a 100Hz sine already looks slightly distorted. 1kHz looks clean.  I will work out this model as far as possible.

I'm not sure what the impedance of an EF86 strapped for triode is off hand, but 200H seems instinctively big enough to not cause too many problems at the low end but?  I tend not to trust distortion results from a sim myself and it's very possible you may not need to worry about it in real life. 

With respect, I assume the Vacuvox U23 design mainly allows for settings on the front panel that were originally on the inside and not needed at hand in broadcast use.

Ah OK, good to know, makes sense.


My plan is to use Edcore TXs and Hammond chokes. I will build a first version with turret boards and point to point wiring. As soon as I get it running I will layout a pcb.


Sounds good.  I don't know where you're located but, if in Europe, there might be some alternatives from Carnhill worth a shot.  I noticed recently there was an inductor in the £20 region that might be a good candidate for anode loading and folks have used their 10K:600 with good results. 

Would you rather skip the input pad and use a 1:1 TX on the input? Could you explain the advantage of having the four secondaries of the interstage TX wired in series rather than having a single winding?


It might be that Rohde & Schwarz were just using the best transformers available to them then...  Could be the mic to p-p grids had  better balance than a line to grids?  I would bet that interstage they used too just required that each secondary be terminated for good response rather than it being a purposeful design decision. 

There's usually a design evolution you can see with these NWDR units and a look at the U73 schematic which came later might give you some clues.  Speaking of which, I think someone on here had the 1:1 transformers  from that unit analysed and copied.  I remember them being quite reasonable in price for what they were, and pretty much spot on to the original coils.

Your spice looks good :)
 
Good evening everyone,

Thanks for your feedback on my ideas. I simulated the frontend of the audio path today and substituted the EAF42 with EF85 / EF89 radio tubes which works fine and maybe I just stick with them.

I will simulate the side chain during the next day or next week and include the features I am wishing for. The attack time is defined by the resistance of the rectifier tube. Any suggestion how to make the attack time variable?

@Winston: I will look at the Carnhill TXs but for some reason I really like the Edcors. Your totally right. The original unit has an output impedance of 30r. You think I should go down that far?

Best regards,
Falk

Ps. Attached is the the i/o plot @100Hz and the spice.


 

Attachments

  • Bildschirmfoto 2020-06-24 um 16.15.37.png
    Bildschirmfoto 2020-06-24 um 16.15.37.png
    534.5 KB · Views: 108
Hey Falk,

I will simulate the side chain during the next day or next week and include the features I am wishing for. The attack time is defined by the resistance of the rectifier tube. Any suggestion how to make the attack time variable?

Switching in different value capacitors on some additional 'Begrenzung' switch  positions would alter the attack time.  The release time resistor in parallel with the cap could be a separate control   
I wasn't sure what 'Ableich' meant so looked it up on Google translate as 'comparison'.  I'm assuming  this means test in this context yes? 


@Winston: I will look at the Carnhill TXs but for some reason I really like the Edcors. Your totally right. The original unit has an output impedance of 30r. You think I should go down that far?




OK yep, Edcors have been used with success on tons of projects. 
30R output z is spot on with other units I've seen so, no surprise there then :D
I don't think it's necessary to worry about it too much. The lowest load that these units saw when in use within old broadcast installations was about half  of the lowest load you're ever going to put on the output, they were designed to drive 300 ohm circuits without loading losses and you'll, maybe, put 600 ohms on it worst case.

Not got my head around all the switch positions yet, there's a fair bit of stuff happening there for sure!

 
@Winston

Switching in different value capacitors on some additional 'Begrenzung' switch  positions would alter the attack time.  The release time resistor in parallel with the cap could be a separate control 
I wasn't sure what 'Ableich' meant so looked it up on Google translate as 'comparison'.  I'm assuming  this means test in this context yes?


Thank you. Sounding great with having a switch for a Cap (Attack) and another for R (Release) in parallel.  Abgleich bypasses the side chain and connects a 12V sine to the CT of the Input TX secondary winding. Allowing for tube matching on the meter which monitors the output in this position.

OK yep, Edcors have been used with success on tons of projects. 
30R output z is spot on with other units I've seen so, no surprise there then :D
I don't think it's necessary to worry about it too much. The lowest load that these units saw when in use within old broadcast installations was about half  of the lowest load you're ever going to put on the output, they were designed to drive 300 ohm circuits without loading losses and you'll, maybe, put 600 ohms on it worst case.


Great, so I will stick with this output Z for the first try.

Btw. really appreciate your well organized style of replying and tried to do so as well.
Best, Falk
 

Abgleich bypasses the side chain and connects a 12V sine to the CT of the Input TX secondary winding. Allowing for tube matching on the meter which monitors the output in this position.


Thanks, figured it was that but hadn't traced through the various bits.


Great, so I will stick with this output Z for the first try. 
Btw. really appreciate your well organized style of replying and tried to do so as well.


Haha, cool thanks.  I'm usually not so organised so, decided it'd be better this way to cut down on clutter.

Does the spice you posted have the correct ratios attached to the input and interstage transformers or are the labels just wrong? 

 
Good Morning,

after a day of optimizing and modifying everything works and all features are implemented such as: Threshold, Attack, Release, Ratio, Gain Makeup. It was quite a bit of pain to get the frequency response right and I am not too experienced with designing circuits but I figured the contributing elements and learned a lot during the past day.  Thank you again.

In the ende the Transformers will be:
Input: 10k:2k4
Interstage: 10k:10k
Output:10k:600r
Sidechain: 10k: 10k

What time constants would you implement? I want to have 5 positions switches. Also for the side chain hipass. What frequencies would you choose?

I attach the frequency response this time. For now I will start to layout a pcb.
Best regards,
Falk

 

Attachments

  • Bildschirmfoto 2020-06-27 um 18.50.27.png
    Bildschirmfoto 2020-06-27 um 18.50.27.png
    163.6 KB · Views: 93
..If it was me, I wouldn't try to decide on the A/R timings before actually having the unit functional - then trying out what was feeling right in real-life situations. then last extrapolating these figures slightly to allow for "extreme-case-use"..

/Jakob E.
 
Get a couple of capacitor and resistor decade boxes.  They allow you to change values on the fly without desoldering.  Hook them up to your timing circuit and tweak til you find combinations you like.
 
Good morning Gentleman,

thank you for your advice. Its highly appreciated. I will get myself those decade boxes and hook them up to the circuit. At the moment I am designing a PCB which I thought was a fun exercise but turns into a time killing monstrous project.

Its a complete new perspective I am gaining by doing all the stuff from scratch (thank God I could copy paste the basic circuit). There is a lot of knowledge needed be gained around the microcosmos of DIY audio that is not to been touched when building kits.

I got somewhat stuck with the trace width and clearance for the plate supply lines (300V 10mA). I decided for 30mills width and 100mills clearance now. I hope this will do the trick.

Have a nice Sunday. Thank you for your advice and thoughts on this build.
Falk
 
Back
Top