K+H UE100 frequency response

[Gold]

[quote author="gyraf"]You have a trimmer for 0dB in bypass - the P01 at "Baustein O+S"..
[/quote]

I think Igloos explanation makes sense. I have trimmed P01 baustein O+S to 0 but no matter how it is set in the Out position it drops 1 dB in the In position. That's no big deal. I can trim it for unity gain in the in position and install a hardwire bypass of the unit.

While everyone is in the mood to answer questions I have another one. The units meet the spec of EQ accuracy of +/- 1dB. Of course I would like to get it better. The trimmers don't have that much range. About a half dB with PD1 on Baustein F. I have found that removing (jumpering) RD4 3,9k I can match channels to within half a dB. I haven't seen any smoke but is this a bad idea?

 

[Gold]

I'll just keep throwing them out there. What would be the approximate input and output impedance of Baustein F for example? I do appreciate the help. These things inspire almost the same awe as my lathe.

 

[gyraf]

Ue-100 "Baustein F" guesstimates

Input impedance: RF24(1M)//RF3(1M) ~ ca. 500KOhm.
Output impedance: Rö12,1 (½ ECC81/12AT7) Cathode follower ~ ca. 3-5KOhm.

Jakob E.

ps: here's a pic of a possible future gyraf: http://www.gyraf.dk/tmp/G-UE100-3U.jpg - what do you think of the alternative user interface..?

 

[Gold]

Thanks Jakob. I had heard you were working on this. I think the user interface looks good. Since the interface of the K+H is like no other EQ I know of it will be confusing no matter what. It has become a given that on an EQ you choose the choose the center frequency and adjust the bandwidth. Not so on the K+H mids. You'll get alot of calls saying that they like the way it sounds but "why doesn't it work like a normal EQ".

If you don't mind talking about it, what have you changed besides the user interface in your version?

Since I'm not happy with the way it sounds flat it looks like my options are either find transformers that I like better or possibly use a solid state input/ output amplifier. Do you have any suggestions?

 

[gyraf]

If you like the sound when bypassed on the unit, but not with the eq circuits "in", The problem could be one of the input/output driver ECC81's in the signal chain (there is one in each filter section, most are still in-circuit even when that band is bypassed) gone tired.

If you don't like the sound even when bypassed, examine the input/output amplifier - the E83F and the ECC81 in the O+S module.

The UE100 specs are very tight indeed - even for a German unit - I do not think it'll be easy - if at all possible - to get even higher precision by changing parts. E.g. the capacitors that select frequencies are all made by paralleled styroflex-capacitors with opposite temperature coefficients (!!!)

I haven't changed much in the original to accomodate the rotary switches - but because step numbers are limited to 11 in the switches I like, this will be the max. steps.. But it'll definitely take a while before I'm ready with a full-working prototype of this one. And it'll be heavy..

Jakob E.

 

[Gold]

Thank you Jakob for your suggestion of looking at tube types in the In/Out Amplifier. I looked around the units I have and the most common tube type was Valvo. The E83F's were Valvo's but two of the ECC81's weren't. So I tried it with the Valvo's and sure enough it sounded great.


[quote author="gyraf"]
The UE100 specs are very tight indeed - even for a German unit - I do not think it'll be easy - if at all possible - to get even higher precision by changing parts.[/quote]

I was able to get the gain of baustein C & D to match more tightly between two units by removing RC4 and RD4 respectively on each unit. Is there any reason not to do this?

 

[andre tchmil]

The UE100 specs are very tight indeed - even for a German unit - I do not think it'll be easy - if at all possible - to get even higher precision by changing parts. E.g. the capacitors that select frequencies are all made by paralleled styroflex-capacitors with opposite temperature coefficients (!!!)



Jakob E.

I'm slowly making progress in restoring a UE100.
So far so good.
Two  of the parallel taped capacitor pairs have broken off wires close to the cap body.
I wonder with what I can replace these in modern times.
Looking for a single Styroflex cap of the same value as given on the schematic. , or going double again ?
Or can I replace them with something like MKT metal film caps ?

 

[jensenmann]

If you want to stay as original as possible then use styroflex in parallel configuration. The values might be hard to come by. But since the type of dielectricum is a part of the sound of a filter circuit it matters.
In case you don´t really care about that, then try to use FKP caps. These are closest to the "ideal" capacitor and can get obtained at 5% tolerance which is closer to the usual 2% or 2,5% tolearance spec of styroflex.
If you want to sink some money in it then mica is an option, depending on the value. These are available at very close tolerances.
I´d avoid any metalized film caps. These have  worse specs than foil/foil caps.

 

[andre tchmil]

Thanks Jens.
I will look into the Styroflex options first, I'm just wondering if I need to keep the parallel configuration with todays ( I hope ) tighter tolerance fabrications.
For the original looks maybe ?

 

[gyraf]

Afaik, the parallelling is about cancelling temperature coefficients (like seen in test equipment) - the two parallel capacitors have coefficients going in opposite directions, i.e. one positive, one negative vs. temperature.

Jakob E.

BTW: I now have a collection of freq.resp. plots for this..

 

[andre tchmil]

Afaik, the parallelling is about cancelling temperature coefficients (like seen in test equipment) - the two parallel capacitors have coefficients going in opposite directions, i.e. one positive, one negative vs. temperature.

Jakob E.

BTW: I now have a collection of freq.resp. plots for this..

so far I can somehow understand the theory,
but does that mean mounting two caps with the same coefficients reversed , i.e. one with the foil in, one with the foil out ?  :-[

 

[abbey road d enfer]

so far I can somehow understand the theory,
but does that mean mounting two caps with the same coefficients reversed , i.e. one with the foil in, one with the foil out ?  :-[

Temperature coefficient is how the capacitance varies with temperature. That's a property of all dielectric materials.
If you parallel one cap that decreases with temperature and one that increases with temperature, the resulting cap will be invariant. At least, that's the idea.
Nothing to see with how the foil is rolled...

 

[andre tchmil]

OK, but how on earth can I order , select , test,  etc these ?

 

[andre tchmil]

anyone ?

 

[gyraf]

I would really REALLY try to save or fix the originals...

Jakob E.

 

[andre tchmil]

I would really REALLY try to save or fix the originals...

Jakob E.

That's a statement I can't deny 
thanks Jakob.

 

[PRR]

> the parallelling is about cancelling temperature coefficients

IMHO.... tempco of any decent dielectric should not be a big deal for a studio EQ. If the studio is too hot or cold, the musicians and engineers will "drift" more than the capacitance.

I do not know why the originals were "doubled". I *suspect* it is a technique to get 2% accuracy from 5+% parts. The larger(?) cap is picked maybe 10%-25% small. A smaller cap is put with it and the total is measured. If way off, pick a different smaller cap. If nearly there, pick a same-number small cap with a different error.

Today, if "originality" is not essential, I would get an assortment of good caps and parallel one two or three until I hit the nominal value. Say you want 10,000pFd. Start with 6,800p and 3,300p. That seems to give 10,100p. However the "10,000p" can be 200p off its number, and the 3,300p may be 100p off. So measure. Take different "3,300p" (which may be 3,200 or 3,400) until the total is right. If high, change to 3,000p. If you come a wee bit off, aim low and add a 68pFd.

All of this may be over-fussy. It is hard to see a need for >5% precision of frequency. If this is an L-C filter, and the coil were perfect(!), the cap could be 10% out and the frequency could be 5% of intended. So this cap-paralleling *may* also be about correcting for the larger slop in coils. Unless CJ was in there winding one turn at a time to get within a mH, then smacking the iron to make it spot-on.

 

[gyraf]

It is  RC filter only, and they are aiming at significantly better than 5%.

Not saying it' s necessary, this was obviously designed by a highly neurotic German.

I have two of these, very different generations, but they match to stupid amount of presicion when set to the same...

Jakob  E.

 

[andre tchmil]

I did my best to repair the original ones, unfortunately the legs were broken off too far inside the capacitor body.
So I've  put in new ones 1% tolerance , should be close enough for my old ears.
Everything in the mid bands is working great again but I'm still having a huge bump when switching from bypass to cut or boost at high boost  settings.
I bypassed ALL other modules , including the output amp.  did a good cleaning on the switches too. replaced all Mullard mustard decoupling caps in the midband module.
Can someone with the same units chime in and let me know if that's normal ?
I'm stuck.

 

[gyraf]

I have the same, so I think the thumping at "bypass-to-high-function-transition" is normal behavior in these...

Jakob E.