ad: k1jek

Running tubes on low voltages by design

Discussion in '"Boat Anchor" & Classic Equipment' started by W5INC, Jun 30, 2019.

ad: L-HROutlet
ad: l-rl
ad: Left-2
ad: MessiPaoloni-1
ad: K5AB-Elect-1
ad: Subscribe
ad: L-MFJ
  1. W5INC

    W5INC XML Subscriber QRZ Page

    Working on a "AM modulator" :D from the late 50s. Some very smart EEs designed this piece and they have their 6SN7 tubes running at low voltages when compared to the tube's max voltages. My 32S-3 Xmtr has 1000V on the plates at idle and the tubes don't seem to mind. 800V is suppose to be the ballpark figure for the 6146s I believe. It was conventional wisdom to run the tubes at high voltages, so the tube will be the most efficient in operation.

    Going thru this piece, I have replaced all PS caps and want to get the voltages to where there are "designed" via the schematic. Max plate voltage for the 6SN7 is 400V. The factory info has the plate voltage of 1 side of the first 6SN7 pin 5 @ 67V and the cathode pin 6 running @ 2.6V. The other side of the 6SN7 has a plate voltage of 180v on pin 2 and a cathode supply of 73V on pin 3. This then gets pumped into a 2nd 6SN7 that is running a plate voltage of 105V on both sides of the triode and a cathode voltage of 3.4v to each side. This then feeds the final PA tubes.

    The EEs that designed this piece are far smarter then I am in these matters, so I am asking the Forum here, why is the first 6SN7 running two different voltages on the plate and cathodes? This design was from back in the late 50s and 60s and these EEs weren't running HV on their tubes. I am guessing that these EEs must have done their homework before going against the conventional thinking at the time? How close is close enough to the factory voltage numbers? 67V on the plate is a low voltage so at 10% tolerance level, would 74v be the max? Detail A is that a 250K pot will be used in place of the 220K fixed resistor coming off of pin 4 when called for.

  2. K1APJ

    K1APJ XML Subscriber QRZ Page

    Well, there are a lot of points to discuss, I'll cover a few.

    It was conventional wisdom to run the tubes at high voltages, so the tube will be the most efficient in operation.

    True, in situations where efficiency matters. But the stages under discussion here are voltage amplifiers, the power output is pretty close to zero, so the efficiency is always pretty close to zero. High supply voltage can cause difficulties when dealing with small signals; the voltages in your drawing are appropriate to the signal levels at hand. Efficiency is not a consideration.

    why is the first 6SN7 running two different voltages on the plate and cathodes?
    The second 6SN7 stage is a configuration called a split-load phase inverter. Note that the cathode and plate resistors are equal, so any cathode to plate current swing produces equal voltage swings of opposite polarity across the two resistors. These two equal voltages are fed to the two halves of the second 6SN7. But, using a very high value cathode resistor (equal to the plate resistor) causes the DC level on the cathode to be very high, much too high for a standard cathode bias arrangement (with the grid at DC ground) to work. So, in this circuit, the grid of the second stage is tied directly to the plate of the first stage, as an expedient to establish a DC level high enough that cathode bias will still work. The plate voltage of the first stage establishes the grid voltage of the second stage.

    Now, the circuit you have questioned is not an original design, it is the input section of the famous Williamson audio amplifier from 1948 or so. The Williamson was one of the first credible "hi-fi" amplifiers and the design was copied far and wide. Innumerable schematics available online.

    How close is close enough to the factory voltage numbers? 67V on the plate is a low voltage so at 10% tolerance level, would 74v be the max?
    Not sure what the exact question is here, if the circuit is built exactly as shown I wouldn't worry if the voltages were within 10 - 20%. Between resistor tolerances and tube tolerances there is a lot of stuff to wiggle.

    Detail A is that a 250K pot will be used in place of the 220K fixed resistor coming off of pin 4 when called for.
    Do you mean that you are using a 250k volume control with the wiper connected to pin 4? instead of all the circuitry associated with the 12AY7? That shouldn't be a problem.
    KM1H, W5INC and N2EY like this.
  3. AC0OB

    AC0OB Subscriber QRZ Page


    The tubes' pin voltages appear to set the tubes into their more linear operating regions on their respective characteristic curves.

    Stage 1 (12AY7A) appears to be two pre-amp stages with Summing afterwards (at the 470k's) for two audio inputs.

    The 500k pots are there to set levels from each input channel. The two channels are summed and passed to the first and second 6SN7 stages which are DC coupled to create a phase splitter for the following stages, and as PJ mentioned, uses simple Resistance-Coupled voltage amplification stages at the third 6SN7 stage for Phase A and B amplification.

    Lower voltages usually mean reduced heating, lower per-stage noise, and longer life.

    For example, why would I place 300 volts on the 12AY7A plate@Ip = 3 mA if that stage is amplifying a 100 mV p-p signal with a stage gain of 10 to get to 1 volt p-p, when the curves show I can get the same gain at a Vp of 75V with a plate current of 1 mA?

    The first stage (12AY7A) appears to have a Vp of 80V@Ip = 0.9 mA and an un-bypassed voltage gain of 16.

    I would not mess with or modify anything in the schematic shown as the circuit appears to be well designed.


    Last edited: Jul 9, 2019
    KC8YLT and W5INC like this.
  4. AC0OB

    AC0OB Subscriber QRZ Page

    What is the control grid bias on your modulator tubes and what types of tubes?

    Is your modulator Class AB2?

    The phase amplifiers should have each a gain of approx. 12. Inject a 1kHz audio signal (through a 0.1 uF cap) at the 510k of each grid with say 100 mV p-p; the output at each 51k plate resistor should should be approx. 1.2 Volts p-p, so 5V p-p at each grid point should get you 60 Vp-p, etc.

  5. W5INC

    W5INC XML Subscriber QRZ Page

    THANK YOU Mr. Andrew KA1APJ and Mr. Pheel AC0OB for the education in these amps.

    I ran across 2 @ 90W Allen organ amps circa 1959, from an estate deal and Allen makes top of the line products. Everything was there for the most part, but 1 had some restoration work partially done and the other unit was pretty much stock, but wires were disconnected from pieces. The partially restored amp had new PS electrolytic capacitors mounted between 2 terminal strips where needed, but no wiring was running to them. Wires from the old setup were disconnected also, so I took my time and got everything correct in that endeavor, but I now need to get the voltages correct for the tube pins. The KT88 plate voltages are dead on @ 581V but the plates of the 6SN7s have 284 volts going to them so I need to get that straightened out. That is why I was asking what % of voltage variation off the schematic number is acceptable. I will get the tube pin voltages super close, since I have to work on them anyways. The folks that designed this piece sure know more then I do and I will bring them back to how the schematic is drawn. Plus, I sure don't want to muck up some nice amps, because I was too pig headed not to ask a question towards the folks who are far more knowledgeable in this field then I am.

    I am thinking of adding 4 small milliamp meters to measure the 4 individual KT88s idling currents. I have -74V going to the control grid right now for the KT88s and the schematic is calling for -35. I will install 4 small bias pots, 1 for each tube, so I will be able to dial in each tube to where it needs to go. Measuring the idling current is still the best and most accurate way to have the amps work in the AB1 environment, I'm guessing? The small milliamp meters are cheap, measure about 1 1/4" in diameter and there is plenty of room to mount them very close to the KT88s to boot. Like working on a 1955 Bel Air car, there is plenty of room to make your own additions.
    Last edited: Jul 9, 2019
  6. W5INC

    W5INC XML Subscriber QRZ Page

    I believe the amp is a class AB1 Mr. Pheel, here is the whole schematic.

  7. AC0OB

    AC0OB Subscriber QRZ Page

    That is a good amp.

    Do all of the 6SN7's have high plate voltages? If they do, someone has messed with their plate resistors and or the filter resistors in the filter string noted in the middle of the schematic.

  8. W5INC

    W5INC XML Subscriber QRZ Page

    Yes Mr. Pheel, someone has been in here with the Golden screwdriver before I got them. All of the plate voltages are at 284. It looks like the HV Xformer on this 1 amp I'm working on now has just been rebuilt or is a new 1. Could be the reason for the higher voltages to the tubes. The KT88 plate voltage is correct, so that is good. There is a different HV supply Xformer in the other amp that I haven't got to as of yet. I did get all of the PS caps from Just Radios last week, who has a very good selection of BA and old radio parts.

    Lets face it at 60+ years old, no telling how many previous owners there have been. So I will get everything put back into stock shape once again, have all voltages right where they need to be and go from there. Once back in stock shape the pieces should be more then good enough for me. I bought some JAN 5R4s from the 50s because of their big glass envelopes along with some OD3s from the same period because of their bigger profile. I will have the amps out where they can be seen, as to me the glowing tube is a work of art onto itself.
  9. AC0OB

    AC0OB Subscriber QRZ Page

    Hi Mike, another suggestion and comment:

    Check the 0D3 regulators. They are two 150V gas regulators in series and a total of 5.3k of resistance between the 580V and the regulators, so there should be about 300 volts on the screens of the KT-88.

    That may be why you are seeing such high voltages on the 6SN7 stages. No regulation would put high voltages in the filter string.

  10. W5INC

    W5INC XML Subscriber QRZ Page

    TNX for the great suggestions Mr. Pheel. When poking around in there checking the voltages the screen was very close to 300V, like 290 or so. But I will sure double check it again and post up my findings. TNX AGN!

Share This Page