Russian rod tubes in a battery all-tube mini amplifier prototype

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi, working on a new subminiature amp!
The circuit is a PP 1j29-b power amp, with 1j24-b preamp and PI.
The OT is a line transformer (10W, 100V, with the 0.625W tap, which should give a 29k primary when using the 4 ohms input with my 8 ohms speaker).
The windings are a little off. I measured 220 vs. 251 ohms from the center tap to the anode connections.

Here is the first draft, with some voltage readings:
n74daap.png


I tested the concept in a turret board, here are some pictures>
7X8EeQcl.jpg

Here you can see the PP stage, two 1J29B (1SH29B) tubes, the 10W transformer, the two batteries and the SMPS.

KU8d5fcl.jpg

Here is the whole thing, with lots of cables, during the debbuging.

a5ia72Sl.jpg

pC9fHR9l.jpg



Please, take a special look at this suppressor driven PI...
I'm not sure if it is working right, it looks kind of out of balance when using my cheap oscilloscope. I don't know how I should design this, other than using the information available here:
Phasendrehstufe mit Verteilungssteuerung

For the high voltage I'm running a MAX1771 SMPS using 2 Li-Ion batteries (which converts from 7.4V to 100.6V). The same batteries are also used for the heaters, where after some tricks I could get something like 32mA at 4.8V (in my case using a 5V regulator).

And a small sound test with a small 4" speaker on the table (no cab, so a little too bright)>
YouTube

It still has some problems, as:
-Blocking distortion with gain maxed
-Motorboating with volume maxed
-A lot of noise (long cables used in this prototype)

To conclude, I think this proves the concept, I just need to fine tune it. I think with a proper board I can test it without all the noise.

If someone has some experience with designing pentode gain stages, or even the PI, please step in!
 
Last edited:
If someone has some experience with designing pentode gain stages, or even the PI, please step in!

Firstly, congrats for doing what some of us :)o) have been gunna do for years.

Secondly, somewhere in my pile of PDFs I'm pretty sure I have some triode wired curves for these tubes. You have the option of moving these to triode mode and dropping the gain. And more easily echoing "normal" amp design practices.

Thirdly, Valve Wizard has a great article on Pentode design. Also in pentode mode you've got buckets of gain and probably don't need anything like the number of stages you're using.

Fourthly, motorboating is inevitable with that design. The golden rule is "no more than two stages on the one cap". See also the usual causes of buzz and much. Put another RC dropper onto the first valve; drop the gain of that stage (by triode strapping) and most of your problems should go away.

Fifth: there are some real challenges with getting DHTs heated and setting the bias. There are some old circuit diagrams around for hand-held radios using these tubes which have some examples. Also have a look what the guys in the "lunatic fringe" of DIY HiFi are doing (e.g. this thread) as they know how to get the best out of DHTs

Finally, blocking distortion indicates grid current. You can either limit the voltage excursion/gain; use ITs; drop your coupling cap values and pad between stages (like a JCM800) or use really big grid stoppers (e.g. per the Soldano)
 
Last edited:
Thank you guys for the comments,

Today I fixed some mistakes and checked the gain of each stage. What is really interesting is that I was heating the 1J29B with only 1,4V at the 2,4V configuration and they were still working. The current was also really low at the heaters. By removing the 5v voltage regulator I got a really strong signal, it is much louder now.

After checking the gain stages I already identified that there is too much gain, it already distorts after the first stage. Only by reducing the volume of the guitar and the gain at the same time I could get some clean.

Another problem is the lack of balance of my PI. One side has a gain of 2,5 against a gain of 6,7 at the other. What I find weird is the fact that the plate is at only 14V while the screen grid is at 48V. I'm missing something here, could be something to do with the grid, which is at 2V.
I could remove one gain stage and make a normal PI, which would have a better balance than this one.

I will read a little more about the tubes. I started using Valvewizards book, but the datasheet that I found is for a screen voltage of 45v and I'm running it at 92V. So I tried the paint_kip to trace the pentode curves and just increased the screen voltage hoping that the resulting curves are something near the real case.

There are some measurements of this tubes over radiomuseum, but again at lower voltages. Interesting is that they show a 1J24b circuit with screen feedback, which should be a triode mode with higher gain (think they managed to obtain up to 100, if I'm not wrong). Maybe I give it a try.
 
I have been trying to analyses it, not as far as actual voltages but just how the damn thing works. I just sort of go around in circles without getting it. Mind you it was only a couple of minutes so far but I can say having the schematic larger doesn't help all that much. ;)
 
Today I changed the PI to an easier one:

syemYNx.png


I adjusted the voltage divider that goes to the second tube with a 500kA pot.
It increased the output considerably.

It was still acting weird though, and at a specific output tube. Turns out I was using the decoupling capacitor at the wrong position. Initially the 10uF capacitor was between the heater center tap and the filament that sets the grid leak resistor. I tested different configurations and the sound was much better when I connected the cap between the center tap of the filament and the negative side of the filament. That also increased the output power.

Now It's as loud as the others subminiature amps that use double triodes in the output stage and require 6v@300mA for the heaters. It changed my voltage readings a lot, since now the amp is drawing more current. It also reduced the blocking distortion.

The amp still sounds as a fuzz with the volume and gain maxed, and I can't get any clean with the gain pot. With the guitar volume at the lowest setting I managed to get some clean, but this way the gain pot is kind of useless, because it only works in with a lower guitar level.

There are still some oscillation problems that produce a tremolo at lower volume settings, which is probably associated with the poor lead dress an lack of filtering. I think I can test it on a PCB now, and see how it turns out without all the cables. I will also add another RC filter to the power supply.
 
I thought that also but what can you do? You have a common supply, not practical to have more than one. I was wondering what cathode voltage you had. Oh wait, directly heated so the cathode voltage will be a fraction of the supply depending on the position. Or would it take the average between the heater voltage and ground potential? On a indirectly heated cathode a suggestion would be to go to cathode bias rather than grid leak. With direct heated? Or go to a higher heater voltage and put dropping resistors inline to give the cathode some more negative bias. But is that what is needed? A scope would be useful to see what is happening to the signal.
 

PRR

Member
Joined 2003
Paid Member
...I just can't get my head around that PI...

If G3 is more positive, more electrons flow to plate, less to G2.

If G3 is more negative, less electrons flow to plate, more are repelled-back to G2.

This is true for any pentode with a separate G3 connection.

The this-way/that-way see-saw is far from perfect. I suspect the 210K resistors are intended to encourage equal outputs via NFB action. Whether that will work, I do not know.
 

Attachments

  • Head-PI.gif
    Head-PI.gif
    11.3 KB · Views: 239
Oh wait, directly heated so the cathode voltage will be a fraction of the supply depending on the position. Or would it take the average between the heater voltage and ground potential?
As I said, step into the next aisle where the DHT HiFi crowd have been playing with this for some years. (e.g.Bartola - much of the stuff dates back more than a decade)

There's a number of solutions but the three easiest are
a) one bias battery for each valve.
b) fixed bias for the valves (then all cathodes are earthed at one end). But you've still got to provide clean heater power
c) constant current sources (from B+) through the DHT and into the Cathode resistor. Current at the cathode is then the sum of the heater, screen and plate currents.


After that and you're into individual heater windings or Coleman regulators
 
Member
Joined 2008
Paid Member
I'm sorry, I really don't get it. Can somebody please explain to me how the collective cathode circuitry as shown is not problematic?
Just for one example, (going from left to right) the cathode in valve three must draw its current from where? Through the cathodes of valves one and two?
I don't understand the bias resistor arrangement either, grounding one end of the cathode, etc.
 
Just for one example, (going from left to right) the cathode in valve three must draw its current from where? Through the cathodes of valves one and two?

Yes, they're all in series, the resistors at the first stages are there to compensate for the extra current that the power tubes require.

Maybe I was too naive thinking that it would work as in indirect heated tubes, since the cathode current is much lower than the filament current itself and the local capacitors would supply the extra current.

I found a similar design here:
Phono Preamplifier with 3a5 Directly Heated Triode

But in this design the batteries are stacked between the filaments, where in my desing the total voltage is applied to the series string of filaments.

I could change it, and the current for the first four stages would come from the battery supplying that part of the circuit. Since the battery supplies 4.2V when full, I could supply the first four stages in series, at only 13mA. They would still be in series, but in a local string, not bound to the power stage, where the higher filament current lies.

I thought that capacitors at the filaments could compensate for the extra current required by the cathode current.
 
Member
Joined 2008
Paid Member
Can you redraw with batteries in? I'm having trouble visualizing what you're describing. and need to see where the bias voltages are coming from.

It also doesn't make sense to me that one end of the filament on valves one and two can be grounded

I'm mostly focused on your complaint about the oscillation/tremolo and wonder if when the cathodes are set up that way, when one cathode pulls current it would deprive another, (what's that expression? robbing Peter to pay Paul.?) causing an oscillation. I don't know this, I just wonder.

I like your thought about the capacitors too, though I wonder if they will actually act like the batteries you hope they'll be. Maybe just making more complex interaction. Again, I really don't know, but I like what you're trying to do and just want to help.
 
Here is only the heater wiring for a better understanding:

KPR30Rp.png


The batteries supply from 4.2V when fresh to 3.0V when empty. Stacked they produce from 8.4V to 6V and this will impact the filament current.

The tubes in the preamp, 1J28B need from 0.9V to 1.4V at 13 mA just for heating. Add the remaining 0.9mA +- 0.45mA of plate current and 0.1mA for the screen.

The power tubes, 1J29B, requires 2.4V at 31mA, but they work (according to my tests, and the tests over radiomuseum) from 1.4V to 2.8V at 31mA for the heaters, plus 5.1mA +- 1.7mA for the plate and 0.5mA for the screen.

In this case, the preamp should require something like 14mA per tube, while the power amp requires approx. 37mA according to the datasheet.

The configuration that I tested, the current design in the picture, uses both batteries stacked to supply approx. 8V, where the power tubes are in series, while the preamp tubes are paralleled in pairs, where a resistor should drop the remaining current so that the whole string has a current of approx. 37mA when considering the plate and screen currents. In my drawing I just considered the filament supply, that's why you will see 31mA and 13mA.

An alternative, considering the link I posted before is to introduce the batteries in the circuit, where the preamp tubes are now at different filament voltages, but still above 0.9V, and the power amp tubes have only 2V available for each, what is still in the 1.4-2.8V range in the datasheet. I also added the caps to have local decoupling and the resistors to drop the necessary voltage.

I'm not sure which design would be better for this application. The alternative version seems interesting because this way I would have some value to play with the grid bias.

For the power stage the bias should be the same, but the tubes are not paralleled. To achieve a close enough grid bias I connected the grid leak resistor of the last tube to the heater string instead of the ground, so the positive bias of the grid with the cathode voltage subtracted is about the same. That would be at 2.4V at the first design or 2V at the second. This could also lead to some feedback, but I'm hoping that the cap to ground would bypass that.

Reading the Radiotron Handbook I found that in DHT the bias should consider the negative side of the filament. Guess that is a bit conservative.

My suggestion would be consider the integral of the filament voltage averaged over the filament, which should result in half of the filament voltage drop.

A positive side of this rod pentodes is that they still work at positive grid bias, up to almost 1V, which helps when the negative side of the filament is also the circuit ground.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.