Weber Copper Cap Rectifiers are not what they seem

Interesting. This is the second time I've heard this (unless you are Joe Rush from another forum).

It is curious. He's right in theory - from the tube spec, you can calculate the simple resistance needed to achieve the same effect. But perhaps the tube spec does not match reality. The Copper Caps may be in spec (for how they were designed), but that's not necessarily the same thing.

Are your current measurements total draw, or just the power tubes? I'm assuming you've measured the total (including the preamp draw and any through bleeder resistors), but if so it is strange that a unit specified for a particular voltage drop at a particular current (the W55/175) does not give those results. I would have thought it was a simple case of working out the resistance value need to give that drop (314 ohms in this case). I think that the Copper Cap contains other components designed to limit the inrush current - perhaps this is the problem, I don't know.

But I would have thought that if you measure your own GZ34, and get Weber to build one that duplicates the voltage drop at a specified current, you should have no problem.

 

OK, I decided to actually measure a real Mullard GZ34. I was very surprised by what I found. I used a cathode-biased GZ34 + 2xEL34 +1x12AX7 amp I'm working on at the moment (Selmer Zodiac 30, power section only).

Unloaded secondary voltage: 555V.

Loaded secondary voltage: 543; rec output voltage 476V. Voltage drop 67V. Total B+ current 137.5mA. Rectifier resistance 67/.1375 = 487 ohms.

Since the current seemed a little high (current without power tubes is 5mA, so each tube is taking 66mA, dissipating 31W at that plate voltage - too much) I decided to rebias slightly.

Now...

Loaded secondary voltage 550; rec output voltage 490. Voltage drop 60V. Total current 110mA. Rectifier resistance 60/.11 = 545 ohms.

This achieved about the right power tube dissipation too, but that's not what's so interesting - it's that the rectifier resistance is LOWER at the higher current. This is the exact opposite of what I expected.

I checked these results with another GZ34 (unlabeled, but I'm pretty sure it's a Mullard) - the same within 1%.

I also tried (using a scrap EL34) running just one power tube. Now, voltage drop is 54V at a current of 81mA - which is why I was using a scrap tube! - but that's a rectifier resistance of 667 ohms, again confirming that the resistance falls as current increases.

This shows that actually Weber (and my own assumption) is wrong - the rectifier tube does not have a fixed resistance. But it varies in an unexpected way.

It's hard to get accurate figures from KLB's measurements without knowing the total current, but I'm going to guess 40mA for the preamp (6 tubes not 1, and one is the reverb driver), which gives, from the 477/428 (49V drop) at 41mA per tube and 40mA for the preamp gives 402 ohms @122mA for the GZ34 - quite a bit lower than what I found, and 475/401 (74V) at 30mA (? - going from "12W" at 401V) about 740 ohms @100mA for the WZ34 - much more.

My guess is that Weber has based the resistance of the WZ34 on that of a real GZ34 at a much lower current draw - about 55mA if the relationship is roughly linear.

hogy, if you're reading this, do you have any comparable figures?

 

John - did you check your pwr tranny secondary voltages? I've been playing around with trying to sim tube rects, and I've found that alot of sag is actually coming from the pwr tranny. Do your tests at stby/quiescence/full tilt boogie. Keep the secondary AC the same for all three and you'll notice the value of resistance for the rect is much lower when it's not factoring in the tranny voltage drop. Also alot of the non linear look to voltage sag comes from the tranny and it's ability to source current.

But I've known Ted for a few years. He is a GREAT guy. I know alot of people in the biz, and Ted and his son T.A. are about the coolest, most down to earth people I've met. We always hook up with a big group of people for dinner at the NAMM shows. The last one we did he took up the tab! One thing about posts like this - tell the man you're posting a somewhat neg review of his product. Even if you are right in everything you say the man deserves a right to see, and make any comments about the topic. Every email story has two sides. PaulC

 

Paul - yes, I did check the PT secondary voltages. I measured the AC then multiplied by sqrt(2) - having first checked that the meter was giving the correct reading like this by temporarily fitting a cap directly to the output of the rec without the rest of the amp connected (I had to physically break the circuit here to measure the current anyway, since the amp has no HT fuse) and measuring the DC on that.

The test voltages I measured were secondary AC x sqrt(2) at the tested current, and first filter cap voltage, so the difference should be purely due to the rectifier.

I did not try at idle and full power - in this particular case it doesn't matter, because this amp is actually true Class A (one of the few...!). It says it's 30W but actually puts out about 25, and will tolerate being biased to give true dual full-waveform amplification (just), although as I said I had to rebias it very slightly to keep within tube dissipation limits; I raised the cathode resistor from 250 to 270 ohms. That's why I picked this amp to test (and because I had it in bits anyway!).

There was however a definite PT sag as the tube current was increased. As the current went from 0 to 137mA the secondary dropped 12V, which is quite a lot, and in a Class AB amp you're definitely going to get some noticeable sag from this.

What I found very interesting is that it appears that the rectifier resistance is not only not constant, but falls with increasing current, and possibly with lower input voltage - although the variation here is much smaller and doesn't seem likely to be the main factor to me. This is actually completely counter to my expectations, and shows that (unlike what I have believed for years) there is some difference between a tube rec and a solid-state one with a series resistance. Hogy in particular will find this amusing since I had an argument with him about it...

I'm still not sure how much practical difference it makes to the sound though. I'd be interested to hear KLB's thoughts on this, since if there isn't a sound difference, why worry if the technical characteristics aren't quite the same? As Weber says, they've sold a lot of Copper Caps and most users appear very happy with them.

 

As I recall, is Ted Weber did solicit actual rectifier data and compared the performance of the Cap to various manufacturers tubes and tube spec data for a range of currents/voltages. I remember the posted results correlated well with both spec data and actual tube performance. Your comparison results with only one rectifier tube do not mean the entire Copper Cap product line is faulty. I have subbed copper caps in numerous amps equipped with various tube rectifiers of different mfg origin. I have found the cap to produce values both lower and higher and on rare occasion the same as the tubes they replaced.

It is possible that Mullards have lower forward resistance than spec (giving headroom, clarity etc and adding to the sonic goodness “reputation”). I have pulled a couple of old Mullards and found much smaller voltage differences than you saw (<=10) when subbing in Caps. I tend to think that the WZ Copper Cap is probably performing pretty close to spec for a GZ rectifier and I tend to believe the Cu Cap befiore I believe the tube in this case. Of course there is always a chance the copper cap you received is not working properly. I’ve used a bunch of these things and only once have I ever heard any difference and it was a slight improvement. The tube I removed was one of those fake GZ34s with low output. The amp tightened up slightly with the cap. Heat given off by the Cap is no different from heat given off by a tube with the notable exception of the filament heat. The real issue is the filament supply. By eliminating the filament a significant load is taken off the tranny. That is the heat benefit that is gained, not a change in heat given off by the rectifier.

As far as the speaker comments, it is kind of like climbing out of a sports car and complaining because the steering is stiff and quick and the ride is rough. Weber makes a high performance speaker with tight tolerances designed to respond to the amp (which isn't always a great thing admittedly). If you read the info that comes with each new speaker, Weber explains that tolerances are very close and careful handling is required to avoid rubs. If you want a less responsive speaker that will not produce cone cry, they can add doping to prevent or minimize it. I have had a couple of rubs on Weber speakers too. One I bought used and apparently it was damaged in shipping from the owner. The other one developed a rub in the amp while playing. Even though I wasn’t running it hard it got a rub. I attributed it to the tight tolerances. Also care must be taken to tighten the mounting screws to avoid twisting the frame. Based on that experience, compared to other mfg speakers I have found Weber’s more sensitive to rubs. That can be a pain but the tradeoff is improved efficiency and response.

As far as John’s analysis I’ll have to think about it and I probably won’t be smart enough to figure out exactly what he is seeing. However looking at the at the tube data sheet, I find the current/voltage charts are non-linear. The highest amount of non-linearity is at the low (<70 ma range). As an example (for a given set of transformer voltage and resistance conditions) increasing the current output from 175 to 200 ma produces a 11 v drop. Increasing the current from 25 to 50 ma produces a 20v drop. If I'm thinking straight it sounds like a higher forward resistance at low currents just like you saw. And the differences seen between the Mullard and the Cap are much more dramatic than the non-linearity could explain. However, I wouldn’t speculate that Mr. Weber was unaware of this or didn’t take it into consideration in his design.

hunter

 

As John said, if you are really interested in getting a Cu Cap to work for you, you can spec the voltage drop you want (just be sure you really know what your total current draw is). In your case you could use your Mullard data as your benchmark. I don't even think there is added cost to do this. I think a variable version is available or in the works that would allow you to sdjust your plate voltage to suit your taste. That is if you really want to use one. Of course if you can hear the high frequency distortion generated by the solid state components in your power supply none of this matters i guess. Fortunately for me i can't hear it.

Even eating shipping you can experiment and buy a boatload of Cu Caps for the cost of a single Mullard and the Caps never wear out. FWIW the only time I have had an amp of mine go down on me at a gig in a lot of years, knock on wood here, was a rectifier failure (and yes it was a Mullard). I always carried a little solid state plug in for emergencies but no more (well it is still stuck in my gig bag but I won't be needing it). My experience is the Cu Caps perform as advertised.

hunter

 

Ahh... why didn't I spot this earlier? KLB, it's no wonder you got different sounds from the GZ34 and the WZ34 - you're running the 6L6s under wildly different conditions.

You set them by power dissipation, which is not at all an accurate way of biasing a tube (it's only really useful as a safety check). By setting the idle current at 46mA with lower plate voltage, this is FAR 'hotter' in terms of operating condition (even though the dissipation is the same) than 41mA with higher plate voltage. This alone could explain the swampy, mushy sound. I should have realised this sooner since I dislike running tubes hot for this exact reason - I prefer the cleaner, bolder sound when they are as cold as reasonable without introducing crossover distortion. This is actually how I set bias - bias cold until distortion occurs, then back off until the amp sounds good. I only check the current (and calculate the power) to make sure it's in the safe region - I never 'set' by current.

Setting by current can also be misleading - again, it isn't accurate because you need to know the other operating conditions. To make a true comparison between the recs, you need to set the power tubes by operating point, ie where the waveform clips relative to the input signal. With slightly lower plate voltage, this might even occur at lower tube current, not higher. This in turn will actually raise the plate voltage a little.

If you do that, you may find the Copper Cap is closer sounding to the GZ34 than you think. It's obviously not going to be precisely the same, but as has already been said there may be some variation in GZ34s.

I'm not totally sure about the switching transients in solid-state diodes. I've heard alot about this but I'm not totally convinced - the filtering should take care of it. It may be necessary to fit a smaller-value non-electrolytic cap in parallel with the main filter cap though, since electrolytics have relatively high internal resistance and so can't suppress very high frequencies well. You see this in hi-fi amps quite often.

 

good discussion, thanks to all.

unquiet wrote: "I did not try at idle and full power - in this particular case it doesn't matter, because this amp is actually true Class A (one of the few...!). It says it's 30W but actually puts out about 25, and will tolerate being biased to give true dual full-waveform amplification (just), although as I said I had to rebias it very slightly to keep within tube dissipation limits; I raised the cathode resistor from 250 to 270 ohms."

I'm blown away by the revelation that the Zodiac is true class A. Is this the case with the stock wiring, or only after you changed out the cathode bias resistor ? It's a shared resistor for both power tubes, right ?
I knew that it put out low power for a pair of EL34s but I thought that was only due to the tubes being cathode biased. A Zodiac 30 is my #1. I have lately been running it with a GZ37 instead of a GZ34 to drop the plate voltage a little, so I can more safely run Siemens (RFT) EL34s, which I like for their tone. I know it works the power tranny a little harder, but it's a big Partridge, I figure it can more than take it.

 

I think it is a true Class A. Whether it was designed like that on purpose I don't know. If so they must have deliberately ignored the power rating for EL34s.

Certainly with the original 250-ohm cathode resistor (which is common to both tubes), the tube current was independent of signal level right up to clipping, showing that the full waveform is being amplified completely by each half of the push-pull - which is the definition of Class A (I don't have a scope at the moment to really check... it died a while ago and I haven't got around to finding another... I really should). But at this it was stressing the tubes too much really - 31W is well over the max, and both tubes were showing some plate glow. Rebiased with the 270-ohm resistor, it was still very close - probably still just on the edge of Class A.

I do think these are great and very under-rated amps. For some reason they had a bad reputation by the late 60s - I really don't know why, as they are very well-made, reliable and IMO excellent sounding. The "Croc-skin" ones are also so cool-looking, but maybe not by the tastes of the later 60s... likewise probably tastes in tone had changed a lot.

The ones that followed, that looked and sounded more like Fenders - the Treble & Bass 50 in particular, which is a very close copy of the BF Bassman but with EL34s, is an incredible-sounding amp - were already starting to show evidence of the styling crimes that ended in the T&B series with the full-height aluminum panels by the early 70s. These are still solid and good-sounding, but just look so awful that no-one wants to know. Over here you can still pick one up in fully working order for £50 or less, usually including a full set of Mullards...