Leslie 21H Zener Diode mod (trying a trick from the late Harvey Olsen)

[dbsoundman]

Hi all, I recently picked up a Leslie 21H amplifier on eBay initially with the intention of using it in a Leslie 45. As the 45 doesn't use a field coil speaker, I needed to find a way to replace the field coil resistance. Initially, I used a 50W 2.7k resistor, but I found that it got REALLY hot in operation; like, uncomfortably hot.

Convinced that there was a better solution, I continued searching the internet, and turned up this old topic where Harvey Olsen suggested using a 39V 5W zener diode in place of the field coil. Here's the relevant passage:

The original speaker field coil also acts as a voltage
regulator of sorts by loading down the power supply. Without the coil in the
circuit, the output screen voltage tends to climb. This causes the output stage to
draw more plate current than it should.

In both 6V6 and 6L6 amps the output screen grids should be 39 volts DC
*under* the plates. The easiest way to accomplish this today is to wire a 5 watt,
39 volt zener in series with the 400 ohm power resistor in the B+ circuit. The
zener is installed after the resistor but before the DC takeoff for output
screens. The cathode (banded) end of the zener diode goes toward the screens.

So, I bought the parts and tried it out. My first attempt was an epic failure because I totally misinterpreted his instructions. After replacing the .25 amp fuse and installing a fresh 39V zener diode the *right* way, I tried it out again. I could see the screen voltage rising to a good level as it was warming up, but I also saw the magic smoke coming out of the zener diode, so I quickly powered down, and now I'm here.

Here's the 21H schematic for reference:


As I read it, the zener diode should basically be connected where the field coil was installed before, with the band pointing toward the line labeled "295V" on the schematic, and the other end connected to the line labeled "19V". Apparently, this is either not correct, or this technique just doesn't work for some reason.

Any tips or thoughts? Has anyone actually tried this out?

Thanks!

[dbsoundman]

Posted this on another forum and was told that the zener diode should be connected in series with the 400 ohm resistor; the screens then connect to the banded end of the diode. The other connection for the former field coil is then not used (the one that goes to the "19V" line).

I'm going to resolder things later today and will post updates.

[David Anderson]

As I read it, the zener diode should basically be connected where the field coil was installed before, with the band pointing toward the line labeled "295V" on the schematic, and the other end connected to the line labeled "19V". Apparently, this is either not correct, or this technique just doesn't work for some reason.

As you've discovered by trial and error, connecting that Zener diode the way you did it was not what Harvey had in mind. Given where he posted, Harvey probably assumed he was writing for an audience that would understand his idea without a lot of elaboration. Putting a 39V 5W Zener in place of the field coil is a recipe for disaster.

You might want to read some electronics tutorials online about Zener diodes. I think that would clarify things for you. Once you understand the concept, the application should be clear. Pay close attention to the polarity of the Zener diode. If it's the wrong way, you'll just get a standard 0.7V diode drop, not 39V.

https://www.electronics-tutorials.ws/diode/diode_7.html

[dbsoundman]

As you've discovered by trial and error, connecting that Zener diode the way you did it was not what Harvey had in mind. Given where he posted, Harvey probably assumed he was writing for an audience that would understand his idea without a lot of elaboration. Putting a 39V 5W Zener in place of the field coil is a recipe for disaster.

You might want to read some electronics tutorials online about Zener diodes. I think that would clarify things for you. Once you understand the concept, the application should be clear. Pay close attention to the polarity of the Zener diode. If it's the wrong way, you'll just get a standard 0.7V diode drop, not 39V.

https://www.electronics-tutorials.ws/diode/diode_7.html

Thanks David. I'm generally pretty keen with electronics, but they don't teach you all the practical applications in college when you get your EE degree! In particular I didn't have much experience with zeners. I did some reading but was misled because most applications use it as a voltage regulator, where the load is connected in parallel with the zener; this is where I made my mistake. I have since corrected my circuit and connected the zener in series with the 400 ohm resistor, in a reverse-biased configuration. Harvey did say that the banded end should point "toward" the screens, but I found that to be incorrect; the banded end points toward the 400 ohm resistor; the other end of the zener connects to the screens. Once I got that all straightened out everything works well!

One other point that was brought up in another discussion on this topic was the current available to the screens with the 400 ohm resistor. It was suggested that I might need more resistance than 400 ohms to limit the screen current. While I think this may be technically true, I also see that many Fender amplifier designs had a very small or no screen resistor, and given the generous design tolerances of the Leslie amp I'm questioning whether or not I really need to fiddle around with different screen resistor values. Any thoughts?

[David Anderson]

One other point that was brought up in another discussion on this topic was the current available to the screens with the 400 ohm resistor. It was suggested that I might need more resistance than 400 ohms to limit the screen current. While I think this may be technically true, I also see that many Fender amplifier designs had a very small or no screen resistor, and given the generous design tolerances of the Leslie amp I'm questioning whether or not I really need to fiddle around with different screen resistor values. Any thoughts?

Screen resistors are a topic of contention in guitar amp tech circles. It is true that many tube amps have worked fine for decades with the screens connected directly to a filter capacitor node. In amps driven into heavy output stage distortion, screen current can increase dramatically, resulting in screen failures, particularly in amps with higher screen voltages. Fender, for example, routinely exceeded tube datasheet voltages. 21H amps are pretty conservative in comparison.

In stock form the 21H amps have no filter capacitor node on the screens. I think adding a filter capacitor here is an improvement. You might also put a cathode bypass capacitor in parallel with the cathode resistor. The speaker field coil provides some shunt regulation of the cathode voltage when it's in-circuit, but, without it, you lose that regulation.

If voltages are running high with the modification, JJ 6V6s are a sort of 6V6/6L6 hybrid and can tolerate higher plate voltages and dissipations than genuine 6V6s.

So you have an EE degree? It's interesting. With the massive amount of material that has to be covered in EE programs, I understand that little attention is paid to tube/analog circuits these days. My grandfather studied EE at Clemson in the 1920s, so he taught me this stuff.

[dbsoundman]

Screen resistors are a topic of contention in guitar amp tech circles. It is true that many tube amps have worked fine for decades with the screens connected directly to a filter capacitor node. In amps driven into heavy output stage distortion, screen current can increase dramatically, resulting in screen failures, particularly in amps with higher screen voltages. Fender, for example, routinely exceeded tube datasheet voltages. 21H amps are pretty conservative in comparison.

In stock form the 21H amps have no filter capacitor node on the screens. I think adding a filter capacitor here is an improvement. You might also put a cathode bypass capacitor in parallel with the cathode resistor. The speaker field coil provides some shunt regulation of the cathode voltage when it's in-circuit, but, without it, you lose that regulation.

If voltages are running high with the modification, JJ 6V6s are a sort of 6V6/6L6 hybrid and can tolerate higher plate voltages and dissipations than genuine 6V6s.

So you have an EE degree? It's interesting. With the massive amount of material that has to be covered in EE programs, I understand that little attention is paid to tube/analog circuits these days. My grandfather studied EE at Clemson in the 1920s, so he taught me this stuff.

I have certainly noticed the hot debate on screen resistors, hence my general lack of direction on the topic. I'm not really inclined to change what I've done on my amp so far as it works fine and I really don't think I'm going to get into a position where I'm driving it way into distortion; at that point I would have launched the voice coils out of the stock speakers anyway. I suppose I could try a higher screen resistor; I'm thinking 1-2k would be more appropriate, but once again, Fender has been building amps since the 50s with 470 ohm resistors on the screens of 6V6s without major issues.

I didn't think of a screen cap, that could theoretically help with noise introduced by the zener diode as well. I did follow Harvey's advice and changed the cathode bias resistor to 150 ohms from 75 ohms and added a 220 uF bypass cap, which worked beautifully; I get a nice ~23V bias voltage now.

The biggest change I've noticed since making these modifications is an increase in B+ voltage, no doubt due to the lighter load on the rectifier tube. (I should note that my 21H amp, which is the 20W version with 4x 6V6 tubes also came with a 5U4 rectifier which seems to be original.) I'm now seeing about 390V B+, so my screen voltage is close to 350V. Again, Fender amps with 6V6s operate in this range as well so I'm not too concerned, but it is interesting to see how the whole system changes.

Back when I was studying for my EE I think I read/researched more about vacuum tubes than I did transistors, but then I graduated and life got in the way so I'm getting back into the swing of things. My day job doesn't have me designing circuits so I'm still trying to get my mind back in gear.

[David Anderson]

If you look at Leslie 122 and 147 amps (or any of them, really), you'll notice that Leslie never used individual screen resistors at all.

Original datasheet values for 6V6s specify 315V max for screens and 350V for plates, keeping in mind that plate/screen voltages are measured from the tube's cathode, not ground, so you're not exceeding the datasheet values by very much.

One thing to keep in mind is that Zener diodes tend to fail shorted in applications like this. Leslie found this out in the 122A when they substituted a single 105V 5W Zener diode for the 0C3 tube. Over time, these have often failed, and Leslie started using two 5W Zeners in series, which is a lot more reliable.

How hot is the 6SN7 running with this change?

[dbsoundman]

If you look at Leslie 122 and 147 amps (or any of them, really), you'll notice that Leslie never used individual screen resistors at all.

Original datasheet values for 6V6s specify 315V max for screens and 350V for plates, keeping in mind that plate/screen voltages are measured from the tube's cathode, not ground, so you're not exceeding the datasheet values by very much.

One thing to keep in mind is that Zener diodes tend to fail shorted in applications like this. Leslie found this out in the 122A when they substituted a single 105V 5W Zener diode for the 0C3 tube. Over time, these have often failed, and Leslie started using two 5W Zeners in series, which is a lot more reliable.

How hot is the 6SN7 running with this change?

Checked the 6SN7 voltage this weekend, unfortunately I didn't write it down when I measured it, but it was within an acceptable range. I want to say the plate voltage was closer to 200V than the specified 110V, somewhere close to 175V if I recall.

I also measured the 6V6 plate voltages with reference to the cathode instead of ground and found the voltages to be more in line with the specs as well, so everything is looking good there.

Regarding the zener diodes, I suppose I could eventually look for some 20V zeners instead of the single 39V that I'm using now, but as I don't have any others on hand right now I'll just have to see how it goes. Two 39V zeners in series would give me too much voltage drop so I don't want to go that route for now. I did replace the soldered 1/4 amp fuse in the screen ciruit for a regular AGX fuse in a fuse holder, so if I happen to have a shorting issue I can more easily replace that fuse at least.

[David Anderson]

I think the quarter-amp fuse was mainly for the field-coil in the event of a short in the wiring. 250mA is still a lot in terms of a screen short. 100mA might be more appropriate there.

You might install a dropping resistor to get the 6SN7 stage closer to specified voltage, anything in the 120V to 130V range. That driver stage can supply plenty of clean voltage swing to drive the output tubes. There's no reason to push it harder.

In the 122A, the Zeners are dropping 105V, so there's more strain on them. Just keep in mind that Zeners tend to fail shorted.

[dbsoundman]

I think the quarter-amp fuse was mainly for the field-coil in the event of a short in the wiring. 250mA is still a lot in terms of a screen short. 100mA might be more appropriate there.

You might install a dropping resistor to get the 6SN7 stage closer to specified voltage, anything in the 120V to 130V range. That driver stage can supply plenty of clean voltage swing to drive the output tubes. There's no reason to push it harder.

In the 122A, the Zeners are dropping 105V, so there's more strain on them. Just keep in mind that Zeners tend to fail shorted.

Good point on the fuses, I need to order some more anyway; in the experimental process of getting this wired correctly I went through a couple more fuses than I'd care to admit! I'll go for 1/10 A instead of 1/4.

[dbsoundman]

Just wanted to note for posterity that this technique does seem to work quite nicely! I did a test-fire of the amp in a Leslie 45 cabinet last night and it sounds great. I wouldn't say it's any louder; the 21H just can't keep up with the later models volume-wise, but it's perfect for me; I can crank it up to 10 and get a nice bit of overdrive without blowing out my eardrums.

I'll post a separate thread on what I did and how I did it in the coming week.