• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

Tubelab Universal Driver Board, 2015 version

I will be back on Monday and I'm going to plug it in!

With the -8 F degree weather we had yesterday there were the usual diversions....my car wouldn't start (OK I'm not going anywhere), Sherri's car had two low tires (easy to fix), my daughter's car wont start.....OK, now I need to jump start my car to go jump start her car, which had a permanently dead battery....so off to Walmart we go (nobody else is open).

But even with the diversions and 3 wiring mistakes, I plugged it in last night at about 9:30 PM.

So what happens when you fire up an amp with the control grid of some Chinese KT88's grounded (0 volts) and the screen grids left unconnected?

In theory, nothing should happen, but modern production tubes often have a less than perfect vacuum, so gas contamination could create a path for current flow.......Well, both of my 15 year old Chinese tubes drew exactly 4 milliamps. No KABOOM!

After fixing another wiring error, the amp was working. With both channels driven by the same signal generator I couldn't get over 50 watts per channel......DUH, my 300 mA power supply didn't like making 600 mA! I hooked up the big one. Both channels are almost perfectly matched. The scope traces overlap from no power to 75 watts where one clips about a watt before the other.

I left it at 75 watts for about 20 minutes while I put my camera on a tripod and took the usual shot in the dark pictures. No glow, no problem, just power. Tube current was 150 mA each.

Today, a few measurements, then SPEAKERS!!!!!!
 

Attachments

  • P1000455_x.jpg
    P1000455_x.jpg
    886.3 KB · Views: 825
  • P1000456_x.jpg
    P1000456_x.jpg
    616.2 KB · Views: 803
  • P1000457_x.jpg
    P1000457_x.jpg
    673 KB · Views: 775
  • P1000458_x.jpg
    P1000458_x.jpg
    650 KB · Views: 768
  • P1000462_x.jpg
    P1000462_x.jpg
    873.3 KB · Views: 769
  • P1000467light_x.jpg
    P1000467light_x.jpg
    750.8 KB · Views: 394
I did a quick frequency response test at 50 watts through the Edcor OPT's. As expected the bass response was outstanding. The lower 3db point was beyond my range of measurement. The first hint of core saturation starts at 10 Hz. The upper 3 dB point is 27 KHz. The amp is 1.27 dB down at 20 KHz. This is with no global feedback around the OPT.

Next I got out my DIY "blow proof" speakers, and took them apart to tighten up all the drivers. These speakers were previously used to test a 600 watt class D amp, which shook the 6 inch drivers loose. For cheap ($20) drivers these things sound pretty good and will eat my guitar playing at 125 watts per cabinet! I connected up the speakers and drove the amp with an iPAD.

I wanted to find out just how much power supply would be needed in a real world amp. I already know that to play 75 WPC sine waves for a week straight would require 450 volts at 600 mA, but I don't need to do that. How much current is needed for loud music that just touches clipping?

So I cranked on this thing for a couple of hours playing all sorts of music at full tilt. I could see clipping on the scope several times, but never saw the current meter go above 450 mA. Much of the current for transients comes from the big poly cap between the output tubes. Even with loud music cranked, the average on the meter hovers between 300 and 350 mA.

This amp feels as loud as the 300 WPC @10% THD class D amp I last hooked to these speakers. I think that once you have those little cones moving at X-max, any more power doesn't make more loud, it just makes more noise (distortion).

I made a YouTube video of a test with some Trans-Siberian Orchestra music, cranked LOUD! (saw them in concert last weekend for the 4th time). Recorded with a $200 phone, it isn't going to win any fidelity contests, but it isn't bad and is in 1080. The humming sound heard is the fan in the big power supply. I'll make an new video when I build a real power supply. The video is here:

YouTube

Next experiment......7403 tubes. I can feed them more voltage, they are rated for 850 volts in triode!
 
Love the video and the choice of holiday music!

I have seen TSO in concert 4 times, counting our road trip to Columbus Ohio last weekend.

I added the necessary plate caps and plugged in some 7403's. I was met by a 200 watt power oscillator operating in the low VHF spectrum. It took some rerouting of wires, and a dozen industrial strength magic beads to tame that beast. Not the 3 big ferrite beads on each plate lead. All the grids already had the small sized beads right at the socket pins.

I can get 100 watts per channel in triode mode but it takes about 520 volts and 700 or so milliamps to get there. More experimenting tomorrow, and maybe I'll hook the speakers back up.
 

Attachments

  • P1000468_x.jpg
    P1000468_x.jpg
    918.8 KB · Views: 439
  • P1000470_x.jpg
    P1000470_x.jpg
    702.5 KB · Views: 391
I spent a good chunk of Tuesday playing with the 7403's, and shelving them for the time being. They have the widest spread in Gm of any batch of tubes that I have played with trying to find 2 pairs that were somewhere near matched from 10 tubes obtained from 3 sources over 10 years shouldn't be that hard, but after several hours I'm still no closer than I was when I started.

It appears that 2 of the tubes may be bad, one for sure as it will go into a runaway condition such that removing all drive and applying -100 volts of bias will not cut the tube off, and the current will keep climbing until I shut the power supply down at 80 mA. This is usually gas, but there was no telltale blue glow.

Either way, pop any 2 good tubes in the amp, set the bias at 45 mA, and let her rip, and you can get 100 watts of triode power at 525 volts. The distortion numbers are nowhere near the GE 6550A's, and 4 out of 4 of the 6550A's read the same current at full power.

I will likely return to the 7403's later on, but unless I can find some matched pairs, I may not use them in my next amp.......'

On to smaller things.....What? yeah little tubes. Yesterday was wasted dealing with health insurance people, a good chunk of today too, but......

I have had this gallon freezer bag full of crusty old 6W6's. I had speculated that these tubes may blow up if allowed to sit idle on 300 volts in triode. I even toasted one of the ugliest tubes a few months ago doing exactly that, so....

I have devised an evil test. I could make up a little amp running 6W6's in triode on 300 or more volts and abuse it for a while to see how they take it.

The picture shows the test amp, and the contents of the gallon bag dumped into a box. These are rather ugly, well used tubes, but it took me about 10 minutes to find two good pairs, and one bad tube out of maybe 7 or 8 tested. I picked the worst of the two pairs (old Sylvanias, I didn't want to fry some vintage RCA's that worked really well) for one channel, and a pair of identical military GE's for the other channel (the control pair).

Both channels hit 3% distortion at around 25 watts (24.82W vs 25.01W) ands their scope traces overlap throughout the entire power range.

The first picture shows the setup. The second picture is the usual shot in the dark, no red glow, lots of good blueness even on old tubes. The third picture shows the power reading and the scope traces (yes both channels are displayed, right on top of each other).

A quick efficiency measurement shows that the output stage consumes 51 watts to make 25 watts of audio, just under 50%. This means that each tube dissipates 13 watts. The spec for audio duty is 10 watts plate, 1.25 watts screen, so we are slightly over spec. One of the crusty old RCA's (later deemed bad) showed red plate in one corner at 25 watts output. It did not bias up like the others and had high distortion, so I believe that it is misaligned from years of banging around.

This testing reveals that the boards work fine down to 300 volts without changing any parts. I will publish the new list as I build out a few more boards an the coming days.

Speaker testing tomorrow.
 

Attachments

  • P1000481_x.jpg
    P1000481_x.jpg
    652.8 KB · Views: 297
  • P1000476light_x.jpg
    P1000476light_x.jpg
    686.3 KB · Views: 350
  • P1000471_x.jpg
    P1000471_x.jpg
    836.5 KB · Views: 378
I wired the 6W6 test amp up to some Cerwin Vega speakers that I got at a thrift store for $20. I used them because they are 8 ohms (my DIY speakers are 4 ohms) which makes things easier because my test loads are also 8 ohms. They are also about 10 db less efficient than the DIY speakers which makes testing the amp at or near full power less annoying for the neighbors.

The 6W6's were wired for triode where they made about 25 watts. Testing with the scope across the speakers revealed some obvious clipping on peaks, some of which was quite audible. So for more power, I wired the amp for UL (just changed the screen grid jumpers) with no other changes. Power went up to about 32 watts, not much of an improvement. The amp sounded good either way, and may be good with more normal speakers or in a smaller room. So where do I go from here.......Into the boxes of crusty old tubes......for some 5881's.

The second picture shows the collection of 5881's (some are marked 6L6WGB, and some have both numbers). These guys have obviously been warm since the bases are discolored on almost all of them. I picked the 4 least toasty looking tubes regardless of brand and plugged them in. The amp was still wired for UL, but I turned the power supply up to 400 volts. I set the bias to an arbitrary 30 mA, then retuned the driver boards for the higher B+. Power output was 45 watts in one channel, and 43 in the other. The amp sounded a bit more dynamic, and I had to pound it pretty hard to see clipping. Except for very loud music, the clipping was not readily audible.

So, while digging through the boxes of crusty old tubes, I saw 4 tubes that would plug right into these sockets, BUT.....I have been telling people that you CAN'T DO THIS for years.....So, I'm going to do it anyway.

Leaving the power supply set at 400 volts, I simply unplugged the 5881's and plugged some 6DQ6 sweep tubes into the sockets. These were old used tubes of 4 different brands, definitely not matched. I set the bias to minimum current (-100 volts) removed the drive signal and flipped the switch......No KABOOM. The maximum screen grid voltage for a 6DQ6 is 200 volts. I'm feeding them 400 volts. Because of this two tubes were drawing 12 and 20 mA already and the others only took a little turn of the bias pot to get to 30 mA.

I have seen sweep tubes runaway rather quickly and sometimes violently under these conditions, and my shiny Edcors are still in there. I'm using the old Fluke power supply since it can't set an OPT on fire like the HP can, but is only rated for 300mA. The amp makes 58 watts on one channel, and 64 in the other. Distortion at 50 watts is in the 1.2% range......Speaker test......

This thing sounds good, so good that I listened to it for several hours. I kept an eye on the idle current which did wander around quite a bit. After an hour of so I noticed that the current went down when the heat pump kicked on, and went down more when it went through a defrost cycle (it's about 5 degrees F outside). Overall the tubes stayed within a milliamp or two 30 mA.

I still will not trust sweep tubes at twice their screen grid rating, but I will experiment further once I swap out the Edcors for some expendable OPT's.

The 3rd picture shows the 6DQ6's in a darkened room. I can't see any obvious parts glowing inside the tubes except for the heater.
 

Attachments

  • P1000500_x.jpg
    P1000500_x.jpg
    766.7 KB · Views: 276
  • P1000498_x.jpg
    P1000498_x.jpg
    760.1 KB · Views: 293
  • P1000486Light_x.jpg
    P1000486Light_x.jpg
    833.7 KB · Views: 359
The Edcors are 3300 ohms. These driver boards are capable of driving just about anything within a few volts of the mosfet supply rails which are +/- 160 volts in this breadboard.

You know, I haven't really thought to put a scope on the grid and see where it's going when I lean on it. I'll do that the next time I have it on. Right now I'm scrounging up all the parts I can find to see how many more of these driver boards that I have stuff to build.

Most sweep tubes don't need or want much positive grid voltage. Previous measurements showed that when driven to clipping the grid was only going 5 to 10 volts positive. 20 to 30 volts on most tubes are OK, and most DHT's will take more, but there is a point of diminishing returns since a lot of them are emission limited, and grid current is electrons that never get to the plate. It is possible for the plate current to actually drop if the grid is pushed too far positive. The 45 tube is specified for 18 watts output in AB2. Most of the 45's that I have scrounged over the years work fine when running at 30 mA or so in class A, but conk out when asked for 100 mA or so pushed hard in AB2.

These 6DQ6's and 6W6GT's are a much older sweep tube design than what I usually play with. Perhaps that's why they lived with 300 volts on the screen. A 6HJ5 or 6LW6 would have blown up or runaway......I have the dead tubes to prove it.

Once I have exhausted my supply (there's 2 more boxes) of tubes that will plug into the sockets directly with it's current wiring, I will rewire it for some bigger, newer octal sweeps including the 6LW6. The LW6 has an odd unique pinout, so it will be one of the last tubes tested. I have some 12 pin compactron sockets that fir this board, and i'll rig up something for some 4 and 5 pin oldie - goldie tubes.....I have a bunch of NOS 307A's and some crusty old 45's.
 
I started thinking and guessed that there was no way these tubes are reaching positive grid voltage. With 400 volts on the screen and 0 or more on G1, they should attempt to pass an amp or two of current......then I thought, in triode, or UL the screen is in the sub 100 volt range when G1 is driven to or above zero.....

Curiosity got the best of me so I walked over and flipped the power switch. I put the load resistors back on and scoped the output. Clipping starts at about 55 watts. I moved the scope to the control grids of one channel and cranked on it until the watt meter read 70 watts. It was probably a near square wave on the output, but the scope was on the grids. I snapped this picture.

The tubes are grossly mismatched, and not even the same brand. One tube does see its grid go to about -3 volts while the other sees -16 volts. (20 volts per line with zero in the center).

I will need to rewire these tubes for pentode mode and feed them 150 volts or so on the screens. My guess is well over 100 watts out without any glow.....
 

Attachments

  • P1000501_x.jpg
    P1000501_x.jpg
    687.5 KB · Views: 793
Last edited:
I rewired the tubes for pure pentode mode with all the screen grids (thin yellow wires) tied together and to a separate power supply set at 150 volts. I kept cranking and tuning to find that the mismatched channel shown in the previous post gets worse the harder you push it. Attempts to get over 80 watts by turning up the main B+ only made more distortion.

I moved all the meters and probes to the other channel to find that it was just quietly humming along at 110 watts and the tubes were relatively well matched having similar cathode currents at 110 watts.

The distortion at 110 watts was 2.6%. This channel had been cranking for 10 minutes or so while I was busy messing with the other one, then for several more minutes while I pushed buttons and took pictures. I turned off the lights and took the usual shot in the dark (then lightened it with a photo editor to make glowing things stand out) nothing cooking at 110 watts output with dumpster tubes. Now ask me again why I like sweep tubes?

Note the scope pictures. The drive on one tube just touches the zero line, and the other tube might go 1/2 to 1 volt positive. That has been my experience with sweep tubes that are not abused, and in decent shape.

The bad channel was pushed about 10 volts positive on one tube as the plate to plate local feedback tried it's best to correct for the sick tube.....I looked at the tube brands, and I can guess which tube is bad......I have never met a decent tube with the Penta brand on it yet. I'm guessing its the bad guy.

I know that I have 4 NOS Sylvania 6DQ6's somewhere. I haven't seen them since I lived in Florida, so I'm not counting on finding them any time soon. I do have 4 NOS 17DQ6's handy. I got them free too!
 

Attachments

  • P1000505Light_x.jpg
    P1000505Light_x.jpg
    772.2 KB · Views: 773
  • P1000509_x.jpg
    P1000509_x.jpg
    923 KB · Views: 784
  • P1000508_x.jpg
    P1000508_x.jpg
    721.9 KB · Views: 777
  • P1000507_x.jpg
    P1000507_x.jpg
    721.5 KB · Views: 789
I went over to the storage trailer and fetched 3 boxes of tubes. One marked 6BQ6GA, and one marked 6AR6 / 6384, both still remain unopened. The third box said 6L6 on it.

I spread it's contents out on the floor to dry out. It was still very cold inside an unheated mobile home since it hit -5F here overnight. The picture shows the contents except for two metal 6L6's that I took to the magnifier looking for type numbers before I found them plainly stamped into the base, and 4 6L6G's that I stuck in the amp.

Yes there are a pair of RCA black plate 6L6GC's to the left of center. There were 9 6L6G's. I took the 4 RCA's with 1940's dates on them and stuck them in the test amp. There are 16 6L6GA's (lower right) and 3 6L6GAY's (lower center). The metal tubes, 6L6G's, 6L6GA's and 6L6GAY's all have the same internal guts. The only differences are the glass and the base material. Lower left are 4 GE 6L6GB's they have the larger plate structure capable of dissipating more power.

So, what happens when I take 4 6L6G's from the 1940's, chosen because they were all RCA's, and stuff them into the amp, which still has 3300 ohm OPT's which are about half what a 6L6 type wants to see. I forgot to check the B+ voltage which was still set over 500 volts since I had sweep tubes in the amp. I flipped the switch and it didn't explode, but setting the bias was touchy. Then I realized that the voltages were all wrong for audio tubes. I set the screen grids to 300 volts and dialed the B+ power supply back to 475 volts.....a bit beyond what a 75 year old tube should have to eat. What happens?

Well 70 watts at 1.6% distortion happens, not on one channel, but both. I let it run for a while and couldn't see anything glowing, but the glass has dark tinting on it. I'm going to test some 6L6GA's since I have a bunch and they're a bit smaller.

Do these tubes go into AB2. You bet. One grid touches +20 volts, the other hits +15. Note a little bit of pointiness on the +20 drive signal. That tube is just bumping into clipping, so the local feedback corrects by boosting the drive just for the period of the clip. Bumping the signal generator up just a nudge raises the distortion up to 2% and creates a little nipple on the drive waveform......forgot to photograph it, but I'm sure I catch it again. I have a bunch of 6L6GA's to sort through.

It looks like I have a bunch of 6AR6's and some 6384's (military 6AR6). These are like a 6L6 on steroids in a small package (like the 5881) so I plan to test them too, but their pinout is not compatible with anything else.
 

Attachments

  • P1000512_x.jpg
    P1000512_x.jpg
    949.7 KB · Views: 328
  • P1000515_x.jpg
    P1000515_x.jpg
    682.9 KB · Views: 300
  • P1000518_x.jpg
    P1000518_x.jpg
    680.1 KB · Views: 185
  • P1000520_x.jpg
    P1000520_x.jpg
    687.5 KB · Views: 198
I have been playing with this board for about a month almost constantly, and I haven't blown anything up.....until now.

You would think that if I was beating on 70+ year old tubes with voltages far beyond the ratings while extracting power outputs in excess of twice the "typical operation" it would have been the tubes, but that does not appear to be the case. In reality I'm still not sure exactly what happened except for parts with chunks blown out of their sides.........

In the last episode I squeezed 70 watts out of some crusty old 6L6G's with no issues. I was cleaning up the lab, and while putting those 6L6's back in the box, I decided to stick the pair of clear glass 6L6GA's into the amp and lean on them to see how much it took to make them glow!

The 6L6GA's behaved differently in the amp. They did not hard clip like the 6L6G's did. They also put out more power. with the same 475 volt supply I could get well into the 80 watt range before the output waveform started flattening on the top, and I got 91 watts at hard clip. I let the amp run for 10 minutes at this level while I set up the camera. Nothing was glowing. The drive waveform was pretty distorted trying to compensate for clipping.

OK, if nothing glows or blows up at 475 volts, turn it up. I kept twisting the knob until I got to 525 volts. I turned up the signal generator until I reached 5% (4.86% really), the power was 105.4 watts. The shot in the dark picture shows no glowing plate. Screen grid current was 7 mA per tube, 2.1 watts nothing glowing there. I turned the signal generator up until the distortion reached 10%, the power meter read 113 watts.

So this can make a lot of power. Immediately after all this abuse, I turned the signal generator down to reach 1 watt of output power. The distortion was 0.416%, it does quiet good too!

I then stuck the pair of Sylvania GAY tubes into the amp and throttled up both channels. I thought to check where the transition from AB1 to AB2 occurs. The drive on the most positive of the 4 mismatched tubes crosses zero at 22 watts. The tube at the other extreme crosses zero at 25 watts. +20 volts is reached at 68 watts, +40 volts comes at 104 watts, +45 volts happens at 113 watts and a big bang happened at 125 watts on the strongest of the two channels, the other was around 120 watts, but it blew up before I could poke it with the scope probe.

Autopsy (really easy this time) revealed a chunk blown out of the side of one of the 1 ohm 2 watt 1% Vishay resistors used for measuring cathode current. I assumed that the associated tube had arced internally, so I slapped in the other GAY tube, a National Union. The amp was alive again.

I proceeded to crank on it again and as I passed the 120 watt mark, BANG! The cathode resistor in the other channel blew. This is the one with the clear tubes that I had been beating on all night. This time I simply replaced the resistor, fired the amp up with the same tubes, pushed it to 100 WPC, and left it there.....no bang.

What's blowing my resistors. These are 1 ohm 2 watt resistors, it should take 1.4 AMPS to blow one of them up, more to blow chunks out of the side. 1.4 amps through a 6L6, not likely, really unlikely without fireworks or damage. That tube still works, and plays nice. I thought that maybe 40+ volts on the grid might be able to do some damage, then I remembered that I have a 100 ohm 1/4 resistor in series with the grid. Getting an amp through that would be impossible. The screen grid supply is incapable of sourcing more than about 350 mA even into a dead short....not there either.

The main B+ supply IS capable of blowing parts in half. It has 1000uF internally across its output, and has blown much bigger parts into black stains on the PC board where a part used to be. Could it do this through a tube????

More experiments to come.....but WITHOUT the Edcor OPT's. If amp sized lightning bolts are coming through the tube, they are also coming through the OPT!
 

Attachments

  • P1000552light_x.jpg
    P1000552light_x.jpg
    718.8 KB · Views: 241
  • P1000545_x.jpg
    P1000545_x.jpg
    681 KB · Views: 183
  • P1000549_x.jpg
    P1000549_x.jpg
    698.4 KB · Views: 233
  • P1000526light_x.jpg
    P1000526light_x.jpg
    781.7 KB · Views: 245
  • P1000528_x.jpg
    P1000528_x.jpg
    596.5 KB · Views: 181
  • P1000527_x.jpg
    P1000527_x.jpg
    687.9 KB · Views: 212
  • P1000553_x.jpg
    P1000553_x.jpg
    665.5 KB · Views: 235
  • P1000567small_x.jpg
    P1000567small_x.jpg
    678 KB · Views: 266
Last edited:
Very interesting how the 6L6 family goes to AB2 so willingly and the sweep tubes seem to resist it.

The 6L6 family is actually specified for AB2 as is the old 45. The "AB2 causes distortion" myth comes from the lack of good driver technology back in those days, not the tubes themselves.

The sweep tubes were designed for very low plate to cathode "saturation" voltage with zero volts on the control grid. This is largely due to the oversized cathode, and a requirement for long life in a TV sweep situation.

A good sweep tube doesn't need positive grid voltage, and the large surface area between these two surfaces will cause a lot of current to flow if there is much positive voltage on the grid.

Many sweep tubes give curves of plate current at zero grid volts for various screen voltages. Late vintage big sweep tubes can pass well over an AMP of plate current without positive G1 voltage.