Lowrey tube amp rectifier

Looking at the picture, it also seems the lone diode is connected to a separate winding of the power transformer, further indicating that it is a separate filament supply.

The 7591XYZ's filaments are clearly directly powered from a winding on the transformer (not seen in this picture, but clear when looking at the bases of the 7591XYZ's). It may be that they are using a 1/4 rectified and filtered supply for the filaments of the 12AX7's for noise reduction (./?)

I'll trace more wires when I get a moment. It should be very easy to see if something is going from here to the filaments of the 12AX7's.
 
Half wave is a single diode. Most likely its a bias supply with that setup. Usually DC heater is full wave rectified just for efficiency reasons. Sometimes bias supplies did double duty as DC heater supplies though.
 
Half wave is a single diode. Most likely its a bias supply with that setup. Usually DC heater is full wave rectified just for efficiency reasons. Sometimes bias supplies did double duty as DC heater supplies though.

Yeah. got that wrong. It also seems connected to the pot, which was not original, so yes, it looks like some sort of bias adjustment.

Speaking of Bias -- It seems that tubes 2 and 4 (numbering them left to right when viewing them from what used to be the connector side, or "front"), emit a very faint red, vertical glow from the center of the plate when viewed from a certain angle.

I am not (at least not directly) seeing the filament. The red glow is viewed 90 degrees from where you can look into the tube and see the filament glow.

It's only visible when you get the room dark enough to also see the blue glow of residual air. It does not depend on volume.

I don't know if this is just particular to individual tubes, or if the plate is actually glowing.

I've read about checking the bias (several techniques), and I've worked with much more voltage than in this thing, but I've never actually done it before in something like this.

I may buy one of those bias checker socket/base things.
 
Here is what I am seeing.

In this picture, I swapped the 2 tubes for that channel. (Left as viewed from the front -- I'll swap them back)

The glow followed the tube, so it must be the tube, not bias.

IMG_20180630_210747_sm.jpg

The 2 small glows at the bottom are the filaments of one of the 12AX7's.

The glow of the other 12AX7 can be seen at the bottom right, in this case directly in front of output tube #3.

Note the 2 spots on the sides of tube #2 where the filament glow can easily be seen on all the tubes.

All the tubes are clocked the same way.
 
well, the 7591XYZ is not really a good sub for the 7591. Its essentially a 5881 with the pins shuffled around. They require more negative bias in order to operate properly, as well as more drive. It would be worth tracing out a schematic or finding an original and comparing to your amp to see what modifications, if any, were made to accommodate those tubes. Possible there were none done and thats why it does that.
 
I'm assuming that this amp was built (re-built, I don't mean originally built) for the 7591XYZ's. Also, my understanding is that if you do not do the mod, and you substitute a 7591ZYX for a 7591, the results are usually more spectacular than a perfectly good sounding amp and a slight red glow.

Also, the glow follows that one tube, not the position that it is plugged into.

I am working on a schematic, and I may be able to compare it to the pictures in the OP.

I've also made a Lexan Bottom with rubber feet. At least I can pick it up now with out worrying that my fingers are going to hit a charged cap.
 
Really rough, traced schematics:

rps20180702_140703.jpg

The Power Supply does not look right at all to me, but I spent more time on that than the actual amplifier schematic below, and I looked at it on 2 separate occasions, and that I what I came up with. I think I just don't know the internal windings of the transformer. There is another tap that is grounded (not shown because I don't know what winding it is on), so maybe that makes it make a little more sense.

Caps 1,2,3, are 150uF 450V.

Also not shown: Another pair of wires coming out of the power transformer. Assuming that it is a single un-used winding (12.6V?)

All tubes, including the 12AX7's have all heaters wired in parallel. There is no other filament supply.


rps20180702_140737.jpg

During this attempt, I found that the ground side of the little capacitor on the screen grids was not connected to anything on one side. The only thing that makes sense, and the only thing it could have possibly reached, was ground, so that is how I show it, and how I (re?)soldered it.

NOT Shown: There is an additional Filter on the B+, so not all B+'s shown are really the same. The additional filter looks like a 10 uF, 450V cap, and a couple of resistors, one of which may actually be a choke. There is a large power resistor from B+ to ground as part of this (not shown) filter. It is mounted to the chassis, and gets quite hot during operation. It seems it would also serve as a bleeder.

I found it interesting that it is grounded cathode, and (thus) the negative grid bias supply, but I guess that was a common design back then.

The capacitor and 56 ohm resistor in the feed back path that is connected to the grounded cathodes also looks a little funny to me, but again, I double checked it. (Note where FB connects in the buffer/inverter)

I wrote 16 ohm on the OPT secondary because that is what I have seen for other Lowery amps. There are no other wires coming out of the OPT's.
 
Last edited:
The .01 and 56 ohm to ground off the output transformer secondary form a Zoebel network. Its for high frequency stability.

Cap to ground off the screens is just the screen decoupling cap. Also fairly typical. Sometimes the power supply daisy chains off it to feed the smaller tubes, sometimes not. It just depends how they designed it.

If the cathodes go straight to ground without a resistor, it doesn't seem like this was modified in the usual manner for the XYZ tube. Additional negative grid supply could be used though. Might be worth measuring that. I'd expect standard 7591 grid supply to be around -17v and a 5881 type to be closer to -25. It also wouldn't be a terrible idea to put say a 10 ohm resistor between the cathodes and ground so you can measure the current through the tubes. If there is only one bias adjustment pot you can get by with a single resistor, but my personal preference is one per-tube so you can see exactly what is going on.

Circuit looks typical for a 7591 amp though. 12ax7 voltage amp, 12ax7 cathodyne inverter, fixed bias, voltage doubler power supply.

The supply is drawn a little oddly but it looks like what I'd expect. Its a full wave voltage doubler. Guessing B+ is somewhere in the 400-450 vdc range? If so the tubes probably should be idling somewhere around 30-35 ma.
 
Thanks for the explanations. I feel a little better about this thing now. (and now that I found/fixed the screen grid decoupling cap).

I have no reason to believe that it was not re-tubed at the same time it was re-built, and the grid bias supply pot has that red marking stuff on it, and it is not moved from the painted position. Like I said, there is evidence to suggest that whoever re-built this thing knew what they were doing, so I am going off of faith that it is OK to keep running it, but I will check the bias as soon as I get a chance.

I probably won't put cathode resistors in it unless it eats tubes.

Thanks for all the help/interest here, and I think I got a great, modernised, P-P amp for way less than most SE amps. It may not have fake wood, or a funky cage, but it looks great sitting on a table, and sounds great to me.

I hope the thread/schematics help/encourage any one wanting to do a similar Hi-Fi conversion.
 
I'd suggest the cathode resistors so you can see how much current the tubes are actually drawing. They also do sell adapters that sit between the tube and the socket to give the same info. Either way you want to know what the tubes are actually doing, the grid voltage alone won't tell the story.
 
OK, here is some data:

Conditions: 8ohm Speakers connected (note that my understanding is that the one an only secondary tap on the OPT's is 16Ohms)
About 10-15 Minutes warm up time of playing music. AC Line voltage (at the power strip): 119.6VAC
Actual test were done with input connected, but no music playing.

Plate voltage (B+), measured before the additional filter, in other words, OPT Center Tap: 421.9VDC

Grid Bias: -37.47VDC at the bias pot, with resulted in about -32VDC actually at the grid,

I don't know the current, since, as mentioned, it does not have cathode resistors. I ordered a bias meter, and socket "pick-offs" for these tubes, and 9 pin bases (for my HK A224) from the Chinese.

So unless I am confused, it looks like he (the re-builder) is going easy on the tubes, right? It runs fine, and I like my Sovteks, so I would probably leave it as is at least until I get/use the bias meter.
 
That might be about right for those tubes, really without being able to measure the current its hard to say for sure what its doing but that doesn't sound unreasonable.
 
My eBay bias checker came in the mail today:


IMG_20180717_181409_sm.jpg


The little bias meter seems to work fine, but I can see an additional advantage of installing the 10 ohm resistors -- It saves the effort and stress of removing/re-installing the tubes numerous times.

Indicated bias currents,
Left to right, when facing power switch side:
56, 76, 51, 68 (mA)

Note that the high one (highest, as they all look high to me) is the one that I previously showed seemed to have minor red plating issues (Now in the position reading 76)

I have not tried moving tubes around to see if values follow tubes or remain with sockets.

According to a spec sheet for a 7591a:

Cathode Current (Max) -- 85mA
Zero Signal Plate Current - Class AB1 Push-Pull @425V is 88 mA

So the numbers I got are way higher than 30-35 ma, but within spec for a 7591a

Is it OK, or is something else going on?

-- Leaky Tubes?
-- Adjust bias voltage more negative?
-- Leaky coupling caps?
-- Other components off value?
 
Thats current per-pair, or 44ma per tube. The plate voltage also factors in here. The plate dissipation is not supposed to be more than 19w. Pdiss is current * plate voltage.

I'd switch tubes #2 and 3 and then adjust things to be not more than 44ma on the highest tube. If you can also measure the plate voltage to make sure plate dissipation is not more than 19w, that would be a good thing too. Possible it needs to be less than 44ma for the tubes to be happy.
 
Plate voltage was measured previously:

"Plate voltage (B+), measured before the additional filter, in other words, OPT Center Tap: 421.9VDC"

Also, I just did a quick "Calibration" of the Bias meter.

First, I determined that the meter is directly connected in series with the (interrupted) Cathode (pin 8, in this case). There is no resistor in the pin/socket thing.

I then connected it to my bench PS, set to a few volts, and current limited to ~20ma. A cheap Digital multi meter set to 200ma was connected in series with pin 8 of the socket/pin of the bias tester. (Both connected across the power supply output -- just the way you normally should not connect an ammeter.)

Since the bias tester only has 1ma resolution, I could only tell it was within about 1ma of the DMM, but it gave me some confidence that the bias tester was not miss-leading me.

Above is nothing about the tubes/amp, just a mini-review for anybody thinking about buying the same bias tester.

---- Back to my situation with this amp:

.076 * 422 = 32.07, so I am way over 19W at idle.

I'll do as you say, and swap the high and low tubes, and then adjust the bias pot until I get < 44ma on the highest tube.

I didn't check the actual output of the (-) supply, so I do not know how low I can adjust it with the pot.

Thanks very much for the help/patience on this.
 
Last edited:
Also, the plate voltage changes as the negative grid voltage is adjusted. For any plate dissipation calculation, I put one meter between plate and cathode, and a second meter across the current-sensing resistor in the cathode.
You can see both readings change as you vary the grid voltage. Keep a calculator at hand, and remember the cathode current includes both plate current and screen current.
 
I didn't open anything in the amp. The open in the cathode connection is inside the socket/pins adapter of the bias meter, (of course it is no-longer when the socket adapters are connected to the meter.)

The "calibration test" of the bias meter was a separate bench test. The amp was not even in the same room.

I just wanted to understand how the bias meter worked, and make sure it was not lying to me.

In other words, the shunt is inside the meter itself, not inside the tube socket adapter.
 
RWood: Yes, I know that increasing plate current will sag will "sag" the plate voltage, and vise versa.

Also, I have been measuring the B+ at the CT of the OPT. It's a safe place to connect the probe. The plate voltage will be slightly less due to the DC resistance of the OPT windings.

So if I watch the B+ and the cathode current, I should be going conservative on the plate dissipation right?

In my case, using my cathode-connected bias meter, and being in the 40 - 60ma current range, about how much of that is grid current? (Speaking per tube, here) 5%? 10%?
 
If I want a quick calculation I will assume 5% of the cathode current is from the screen. If I really want to know, I'll measure the voltage drop across the screen resistor and calculate the screen current, then back that out of the cathode current.

Most of the time, though, and using the two-meter method above (and Pomona Mini-Grabbers to secure the meter leads) I multiply one meter reading by the second one's reading, times .95 to account for the screen. Divide by 1000, if necessary, to get from mA to A. Then with your resulting plate dissipation, divide by the tube maximum amount (such as 19W) to see how close to the target (such as 70% often used for grid-biased amps) it is.

All this take longer to describe than to actually do. And that's good, because with any change in the grid voltage, it needs to be done again!
 
Last edited:
Its screen current that you'd have to worry about, not grid current.

Datasheet typical specs show 88ma per pair plate current and 13ma per pair screen current, so basically 15%. Usually I just ignore the screen current to add in some safety factor.
 
Back
Top Bottom