Super Twin Reverb: HUGE Ck in the First Gain Stage - Diode Bias?

[bob p]

I've got a 1978 Super Twin Reverb that's on the bench for long overdue restoration and PM. It's the 6x6L6 amp that produces 185W RMS. Looking at the schem, this amp is kind of an oddball.

http://www.thevintagesound.com/ffg/s...verb_schem.jpg

This poor amp lived a hard life on the road, got abused in handling, and didn't get repaired when somebody dropped it and broke the reverb tank. It's had some repairs performed by someone with questionable skills. (PT was connected with wire nuts!) For the past decade or two it's been neglected in the corner of a damp garage. Cosmetically it's pretty rough, but electrically it's servicable. I'd like to resurrect it and give it new life as a bass head.

One of the amp's nice features is that it uses cathode follower drive for the 6x6L6 output section. It's different in many respects from the standard SF era amps, which brings up an interesting question.

The first gain stage has a cathode resistor that's 500 uF @ 15V. WOW! The part seems to be a factory substitution for the part spec'd on the schematic, which is 750 uF @ 3V. WOW! That's a HUGE value for a cathode bypass cap. "Normal" BF and SF amps use 25uF @ 25V cap in this location.

So why did Fender use the huge cap?

First, I think that the idea was in part to get unity gain all the way down to DC. Of course, you should be able to accomplish this with a 25 uf cap, like the one used in most BF/SF amps. A theoretical problem in accomplishing this with a 25 uF cap is that low voltage electrolytics tend to have a high ESR, and when operated at voltages below their rating they tend to exhibit low frequency distortion. The textbook cure for this is to use a large cap with a low/appropriate working voltage rating, which I think is what Fender was trying to accomplish. I'm guessing that Fender couldn't source a low-voltage part, so they used something available with a 10X voltage rating in production. Unfortunately, using such big electrolytic caps can impair transient response, causing "sluggishness" or "wooliness" in the tone.

Although it's tempting to just yank the cap and replace it with a 25-25, I would like to optimize the low frequency response of the first gain stage, if for no other reason, than to retain as little resistance to AC as possible. I'd really like to reject as much heater and power supply hum as possible in the first gain stage. To accomplish this, I've thought of ditching the Ck altogether, and biasing the first gain stage using a diode. Theoretically, a diode is advantageous in this regard because it allows a consistant voltage drop across it, which will make the cathode voltage stay constant regardless of anode current. The theory also says that this allows the stage to achieve flat frequency response to DC while avoiding the "sluggishness" that's possible when using a giant value Ck. Converting to diode bias sounds like a win-win solution.

Right now I'm thinking about using a diode to perform the first stage cathode bias, including a snubber cap in parallel with the diode to mask any diode switching noise. I've never done this before, so I'd appreciate any words of wisdom about doing this sort of thing in a Fender circuit. I looked for suitable diodes with a 1.4V voltage drop, but didn't find anything that looked useful. If anyone's done this before, I'd appreciate your help in identifying the right parts for the application.

This Super Twin Reverb led a really hard life on the road. It's pretty rough cosmetically, and it's hard to consider spending the money to restore it to it's native state. Because these amps are so unloved by guitarists, it just doesn't seem worth the effort. Bassists, though, really love the Fender Studio Bass amp, which is essentially the same circuit. So I'm thinking about re-boxing the electronics in a head-style cabinet, and turning the amp into a bass rig.

Any helpful ideas on the huge Ck, or re-biasing with a diode would be greatly appreciated.

[madkatb]

Merlin Blencowe (the Valvewizard) gives a really comprehensible treatment of triode gain stages here:
http://valvewizard1.webs.com/Common_Gain_Stage.pdf
The section regarding bypass caps starts on page 25 and diode biasing is talked about near the bottom of the page.
This link is from his web page here:
How to design valve guitar amplifiers
Both of his books are a great read for anyone interested in working on amps.

[bob p]

Merlin Blencowe (the Valvewizard) gives a really comprehensible treatment of triode gain stages here:
http://valvewizard1.webs.com/Common_Gain_Stage.pdf

Thanks for the link! I've read that chapter, and unforunately it doesn't get me any closer to the answer.

Does anyone have exeperience in doing this? Even if you haven't done the diode bias technique, I'd still be interested in your experience with this circuit. Thanks again.

[Enzo]

Just a thought: do you realize that in the time it took to post the question and read the responses, you could have tacked 10 different values of cap in there and heard first hand what differences it made or did not make to sound and response? Including diodes.


1.4v diodes? You can put more than one regular diode in series to add up the drops, or consider LEDs. They typically have much higher forward drops than rectifier diodes. Different colors tend to have different drops, so blue might be larger drop than red, for example.

I am not sure how cathode bypassing would alter the hum problems.

[bob p]

Gee Enzo, it's not like I come here and ask THAT many stupid questions. It's been a while, at least.

I guess the difference between our approaches comes down to theory vs. practicality. My point in asking the question was to see if other people here had experience, and to see if there's more theoretical knowledge out there than I might learn from. I guess that I'm just one of those guys who would rather derive the best answer from theory and math before grabbing the soldering iron and finding out what gets me close enough, and then compare the outcomes of the theoretical vs. practical approaches.

As it turns out, I was able to locate some diodes with a 0.75 voltage drop in an axial lead configuration. Two of them in series should be just the ticket. They'll get tacked onto the next parts order. One bad thing about LEDs is that regardless of color the voltage drops are just too high to be useful in this aplication.

Thanks for your help.

[Pedro Vecino]

The hum problem in this model can come from several sites but I have seen many times that the ultra linear system complicates the things and the dependence of the matching in the 6L6 is critical together with the symmetry of the bias circuit, thermal stability, etc. The most effective solution is to adjust the bias balance symmetric (reading the voltages on each side) and installing a choke and a 50uF capacitor at its output (this last, away from the 6L6) to power the screen grids and from there, the rest of the preamp.
The original ultra-linear circuit is a nightmare.

[bob p]

The hum problem in this model can come from several sites but I have seen many times that the ultra linear system complicates the things and the dependence of the matching in the 6L6 is critical together with the symmetry of the bias circuit, thermal stability, etc.

I'm not sure that I understand what you mean in referring to how the ultralinear design complicates things, so if you could explain that, I'd really appreciate your help. One thing I can say for sure, the inside of this amp is a rat's nest of wires -- the worst I've ever seen in a Fender amp. The problem is made worse by accumulated dust. The amp is incredibly dirty. I'm tempted to hose it down!

If matching output tubes is critical, that could definitely be contributing to my problem -- the amp came with 5 different types of 6L6 in the output section: a pair of Peavey 6L6, a pair of TNT-branded 6L6, a Fender-branded 6L6, and an RCA blackplate. I got as far as running all of the tubes through a tester last night to verify that they're all "good", though I haven't tried to "match" them. The owner of the amp did not ever bother to buy a matched sextet. It looks like he just threw into the amp whatever loose tubes he could find. I'm sure that a matched sextet would improve the situation. I just haven't gotten far enough into restoring the amp yet to make that cost commitment. Until I do that, I'll try pulling different pairs of power tubes to see if the noise problem improves at all.

The most effective solution is to adjust the bias balance symmetric (reading the voltages on each side) and installing a choke and a 50uF capacitor at its output (this last, away from the 6L6) to power the screen grids and from there, the rest of the preamp.

I'm a little confused by your recommendation. I'm not sure if you are suggesting adding that LC filter stage to the main supply B+ or to the bias circuit?

Are you suggesting converting the amp from UL to pentode operation? If so I'm thinking that the layout would continue to feed the PT center tap from the 500V supply point (right after the standby switch), and that you're recommending that I disconnect the UL taps on the OT, instead feeding the screens with the new LC filter stage, taking the supply off of the right-hand side of the 2700-ohm/10W resistor.


Bringing up a new problem, I am certain that the 35-year old electrolytic caps in this amp have to be contributing to the noise. The electrolytics are ancient. Part of my management plan includes replacing all of the electrolytic caps in the supply rail, and bypassing them. But I don't have the oddball caps on-hand that are needed by the voltage doubler. I'll have to order some parts before I can get started.

Thanks for your helpful ideas.

[Pedro Vecino]

Regardless of the typical problems (aged sockets with poor contacts, ineffective filter capacitors, some disorder in the wiring, 6L6 with different intensities, lack of symmetry in the driver circuit because drifting value in resistors...) hum in this amplifier is is an endless fight. It´s practically impossible to do silent (absolute), and the reason that complicates the things once the other origins has been treated is the ultra-linear circuit. What I did was use a classic pentode circuit with extra choke and filtered output. This makes it much quieter and much less dependent on tube selection (matching). Of course, if you have other problems (typical or new), the first thing is correct them.
Respect to the big cathode electrolytic (750uF) I not find the explanation. Where is the roll-off that occurs? ¿0.3 Hz? I think there is anything to scratch... I use one new 22uF and no perceived difference...

[bob p]

...
What I did was use a classic pentode circuit with extra choke and filtered output. This makes it much quieter and much less dependent on tube selection (matching).

Pentode configurations normally supply B+ to the screen grids from the filter node that's located one step farther down than the B+ feed to the OT center tap.

If I understand, you replaced the 30K-ohm / 20 W resistor with the 50uF / 500V cap, and you replaced the 2700 ohm / 10 W resistor with the choke, and then supplied the screen grids from the node filtered by the new LC filter node, similar to what was done in the older 100W Twin Reverb that also used a voltage doubler. So you replaced the exsisting R-R filter stage with an LC filter stage, so it looks like this?



Or did you just add the new LC stage before the existing R-R stage, and supply the screen grids from the LC node, like this:



I just want to make sure that I follow what you did. The 100W Twin Reverb that uses the voltage doubler and the LC filter doesn't use the huge 30K resistor to ground. It uses a 20uF / 500 V cap in it's place. I'm thinking that what you did probably looks like the first schematic, but I just want to eliminate any ambiguity.

Thanks again.

[bob p]

Respect to the big cathode electrolytic (750uF) I not find the explanation. Where is the roll-off that occurs? ¿0.3 Hz? I think there is anything to scratch... I use one new 22uF and no perceived difference...

The explanation for why manufacturers use such huge caps is on pages 25-28 of this doucment:

http://valvewizard1.webs.com/Common_Gain_Stage.pdf

22uF is probably adequate for the job, but I'd rather find something in a tighter tolerance than what is available in small, low-voltage electrolytics. I'd also like to find a cap with lower ESR and longer life. My preference is to replace the electrolytic cap with something that is less likely to fail over time, like a poly cap of some sort.

I have found that Vishay makes a 33 uF / 10 VDC film capacitor with axial leads that might be a good solution. It certainly overcomes all of the shortcomings of the various types of electrolytics, including the LF distortion and contributing to blocking distortion when the gain stage is overdriven.

[bob p]

I ran across an interesting post on another forum regarding the Fender Studio Bass amp, which uses the same basic topology as the Super Twin Reverb, but omits the reverb and distortion circuits:

diytube.com :: View topic - Replacement Output Transformer for Fender Studio Bass...

Here's what the author had to say. Allegedly his information is based on personal communication with the amp's designer:

I know my response is late -- but based on your original question, I wanted to give you some input about the Hammond transformer you questioned using, that I don't think was addressed in the other responses. I would also like to suggest in the kindest terms, that some of the information was simply wrong.

The 1650T transformer with screen taps would be a marginal transformer choice for the super twin -- and it would require more circuit modification than indicated to make it perform even at a marginal level.

Many people in the guitar amp world mistakenly refer to the Fender tapped screen circuit as an Ultralinear design. It is not, and the taps were not intended, nor are they capable of producing Ultralinear operation. Typical Ultralinear operation for 6L6 class tubes requires a 43% tap (based on turns ratio), which is about what the Hammond transformer provides at 40%. True Ultralinear operation results in much lower distortion, and much lower power output -- about 35 watts is the absolute maximum RMS power output that a pair of 6L6s can give in an Ultralinear design. Therefore, when using the tapped version of the 1650T in the Super Twin, the amp would not produce much more than about 110 watts continuous RMS output. For this transformer to work to it's best ability, the screen taps would need to be taped off, and then have a standard choke type filter system installed for the screen B+ take off. Under these conditions (strict pentode operation), the amplifier would produce about 165 watts continuous RMS output.

As for the actual design of the Fender tapped screen circuit, the purpose of these taps was a cost saving measure on Fender's part, in that they eliminated the need for the choke to provide adequate filtering for the output stage. That was it, and nothing more. This is not an opinion, but a fact that came out of a conversation I had with the circuit's designer. As such, the taps are appropriately placed at only 12.5% of the winding -- enough to provide a small amount of B+ filtering, but hardly enough to reduce power output and cause Ultralinear operation.

Finally, contrary to what was claimed, the Super Twin is quite capable of putting out 200 watts continuous RMS -- if the P-P impedance is correct to allow for it. In this case, the original Super Twin transformer has a P-P impedance of 1500 ohms, which is proper for six 6L6s in Fender's tapped screen design. This is why the Hammond transformer, as a 1900 ohm unit, will not allow full power output to be produced even without screen tapped operation.

When the impedance mismatch, inappropriate tap positioning, and power rating of the transformer are all considered, you can see why I maintain that this transformer would be a marginal choice at best.

Your final decision to operate the unit as is was best. Using the existing transformer into a 4 ohm load will reduce power output somewhat, and is a little harder on the tubes, but it is clearly the lesser of all evils.

It was interesting for me to learn that Fender's circuit isn't actually an UL circuit, and that the reason for the 12.5% screen taps was to eliminate the cost of the filter choke! If this information is accurate, then the idea for switching to an L-C filter stage when converting to pentode operation makes sense.

[Pedro Vecino]

Hello. I followed the second scheme, installing the choke without interfering with the rest of the power line.


I imagined that it was not exactly an ultra-linear transformer with taps to 43% due to loss of power that would entail. I've always considered a pseudo ultra-linear, and indeed the best explanation is to save space and components.
Same problem (hum) I also found in some Twin Reverb 135 watts, having needed to select 6L6 with extreme tolerances and total symmetry between sides to minimize it. In this, probably because he has more tubes the problem is even more serious. With standard Twin Reverbs this problem does not exist and it´s possible to obtain absolute silence without problems.
Regards