• 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.

On going saga with Tubelab SE

I really need some help trying to figure out what is going on with my two Tubelab SE boards. The following is an email I sent to George concerning the first board that was operational until installing the Hammond 270HX PT.

I also have two additional questions. First does anyone know of any better quality higher turn count pots for the Tubelab boards, and second does the B- need to be lowered to the -150V as indicated on the schematic.

Hi George;

Hello from Homer, Alaska. Let me start out with stating right out front that I do not understand electrically everything that is going on with your design, do not have enough knowledge. I have been building speakers, amps, preamps, and a lot of kits or circuit boards projects for over 20 years now. Have only had one problem project (a 832A amp build from schematic) until now. I just do now understand what is going on here.

I purchased two Tubelab SE boards, after completion only one was operational, used the forum but did not get enough information to correct the problem with the second board, will deal with the second at a later date. The first board (components on the reversed side) has not had a problem until yesterday.

The hardware used:

Hammond 270FX: 550VCT @ 173mA; 5V @ 3A; 6.3VCT @ 5A no load voltage = 303VAC, voltage with load = 176VAC

Hammond 270HX: 550VCT @ 200mA; 5VCT @ 3A; 6.3VCT @ 6A no load voltage = 301VAC

OPT: James JS 6113 HS, Electra Print BE5KB, Edcor GXSE10-8-5K

C4 = Obbligate 2 uf

R4 = Triad C-14X

C5 = 2 X ASC 50 mfd 370VAC X386S, for a total of 100uf

Specs with 270FX: B+ = 277VDC @ 28-30mA; B- = -199VDC with 270 ohm R5, with 820 ohm 12W at R5 B- = -180VDC; 175VDC at 417A WE

There was not a problem until I replaced the 270FX with the 270HX PT. I did this in order to use both the 45 and 2A3 on this board (higher 6A 6.3 voltage for the 2A3). Right at startup this first board started acting like the second board, no voltage reading across R29, B+ @ 330V and reading at R18 and R29 not adjustable via trim pots. Removed the 270HX and reinstalled the 270FX.

With the 270FX installed the B+ was higher by 10V to 287VDC, R18 and R29 now became "interconnected", they each effected the others channels voltage, change one side up the other side goes down and .217 became the highest voltage possible, but the amp did play without any apparent problems. After several minutes everything turned to normal and was able to adjust both side to 287VDC @ 29mA. Left the SE running for break in, checking approximately every 30 minutes. At the 2.5 hour check the readings at R18/R29 were back around 21mA and "interconnected" again.

It has been 5 hours since shutdown, restarted the system, specs. are: B+ 283VDC @ 21.7mA (both channels), B- = -184VDC, R18 and R29 both .217vdc, and both 417A's near 175 volts.
System sounded normal.

George as you can see I need help. This is almost the exact same problem that I was having with the second board; on the second board I changed out D2/D3, Q1/Q2, and U2/U3 with no change, absolute no reading across R18 and R29. But, this first board does have reading of .217, will not go any higher, and the pots R12/R23 are "interconnected".

Cordially,
Lee Hankins
[email protected]
 
Some "interconnection" is expected when making bias adjustments. The is due to the B+ sagging under the load of the tube you are adjusting up. It sounds as if your B+ is not staying stable. When you say you measured "287VDC @ 29mA" on both sides, where are you measuring B+?

The best way to help is for you to take voltage measurements when the problem is occurring at the plate, grid and cathode of all tubes and at the source, gate, and drain of both FETs. Also report B+, B-, and the filament supply voltages at the time. From that we might be able to help you further.
 
Hi Russ thank you very much for the response.

The 287VDC @ 29mA is the B+ measurement, B- is -199VDC with stock 270 ohm resistor R5.

Before I installed the 270HX Xfmr, the normal specs with the orginal 270FX xfmr were B+ = 277VDC @ 28-30mA, B- = -199VDC, and 175VDC +- 1V on the 417A's. The amp was broken in for about 10 hours at these specs.

Will get the values you requested shortly.

You are correct that there is always some interconnection between R12 and R23, but now .217 V is the maximum for either side.

The B- at -199V bothers me. In the instructions and schematic it is -150V, and it has to be transformer dependent, design based on 500VCT xfmr. Should I adjust the resistors R5 and or R6 to drop the B- closer to the designed -150 value.

Lee Hankins
 
I'd take a good look at the bias trimpots, R12 and R23. If your B- is -184VDC, you should have -90VDC on one side and -16VDC on the other. With this much range of adjustment, I have a difficult time figuring why you can't dial in the right point to get your tubes biased where you want them. If I hear what you are saying, you've got the trimpot turned all the way over to the -16VDC side. Under those conditions you should be getting much more than 21.7mA through the 45 tube.

Maybe you could check the voltage across R14 and R25, or we can figure it out if you've measured the voltage at the grid of the output tube. I'm guessing you should have somewhere from 90 to 165 volts across it (4.5mA to 8.25mA) depending on where the bias pot is set.
 
The 287VDC @ 29mA is the B+ measurement, B- is -199VDC with stock 270 ohm resistor R5.

But where are you taking your B+ measurement? There are components between the power supply and the plates of the output tubes.

Will get the values you requested shortly.

You are correct that there is always some interconnection between R12 and R23, but now .217 V is the maximum for either side.

Agreed. That's why it is important to see the actual plate voltage (at T2-2/T3-2) and the actual grid voltage (at R33/R34) on the tube. Then we can see why it is not drawing much current through R18/R29.

The B- at -199V bothers me. In the instructions and schematic it is -150V, and it has to be transformer dependent, design based on 500VCT xfmr. Should I adjust the resistors R5 and or R6 to drop the B- closer to the designed -150 value.

It's an indication that not much current is being drawn from the bias supply. It's a "long" RCRC filter that will sag towards 0V as it comes under load. This tells me that the FETs are not flowing as much current as they should. What is the voltage spec on your coupling cap?
 
The B- at -199V bothers me.

I'm really not convinced that is the problem. The trimpots offer such a broad range of adjustment, you should be able to compensate. I don't think the 2SK2700 mosfets care - if I'm reading the datasheet correctly, they can handle up to 900 volts from drain to source.

Can you pull the power tubes (but leave the rectifier tube in) and stick your meter probe in the grid pin of the 45 socket? What voltages do you measure (referenced to ground) at the grid pin as you turn the trimpot from one extreme to another? I'm expecting you should see about -22V to -99V. George actually has you go through this exercise in his checkout procedure:

Next, put in a rectifier tube. Power up the board. After the rectifier tube warms up you should have B+...

Put a meter from the grid of the output tubes to ground. One meter for each tube. Power up the board. Each meter should read a negative voltage. With ONE HAND BEHIND YOUR BACK carefully adjust one of the bias pots (R12 or R23). The negative voltage for the corresponding output tube should change.
 
Last edited:
It's an indication that not much current is being drawn from the bias supply. It's a "long" RCRC filter that will sag towards 0V as it comes under load. This tells me that the FETs are not flowing as much current as they should.

Lee said his power transformer made 303 volts with no load. Multiply by sqrt(2) to get peak rectified DC voltage = 428 volts. The power supply filter has R5 (270 ohm) and R6 (10K ohm). If you pull 22mA across those resistors, you'll drop 229 volts. That brings B- right to 199VDC, just as Lee noted.

I think 22mA is probably about right. You'll draw 0.8mA through each voltage divider (R11/12/13 or R22/23/24). That leaves 10.2mA for each mosfet, which seems about right. I can't find a 2SK2700 for LTspice modeling, but someone suggested the SPA11N60C3 was pretty close. It draws anywhere from 5mA to 9mA, depending on where you set the trimpot.
 
Tubes In: Left Right
278.3 277.4
-60.0 -55.8

Tubes Out:
321 322
-19.6 to -20.25 to
-101.6 -109.3

C9 Cap is .22uf @630VDC

At this point we entered the Twilight Zone. Replaced the 45 tubes and all readings became normal; B+ = 278VDC @28-29mA, then all of a sudden the right channel went to zero, left up to .39V. Then both channels were adjustable to a max of .217V, but the pots became reversed, i.e., turning right increased voltage, left decreased voltage. A few minutes later back to normal with correct rotation of the pots, then right channel went to zero, left up to .39V.

Right pot: left = -16.37 to -88.9
right = -18.49 to -100.1

Left pot: left = -16.5 to -88.4
right = -18.61 to -100.0
 
Everything seems to be OK, another Twilight Zone moment.

I noticed that the CCS IC's sinks had 286 volts on them, thought that this should not be correct. Checked the instructions: "Assemble these in the same manner as before, except that there are no insulators, and no shoulder washer. One heat sink gets both 2SK2700 mosfets. The other gets both of the CCS IC's".

I installed insulators and shoulder washers isolating the CCS IC's from the heat sink. All is now well in OZ. The amp has been playing for over 10 hours with no problems, so far. This really has me stumped, how can the IC's being attached to a heat sink that is not connected to anything else (just soldered to the PCB), cause the IC's unstable behavior. My heat sink is not the standard one from the parts list, it is a solid CU sink that I had to build because DigiKey was out of the called for heat sink, which is now discontinued. They do have an identical replacement.

I would like to thank Russ and Ty for trying to help with my Tublab SE problems. Hope there isn't another chapter in this saga, at least not with this board.

Cordially,
Lee Hankins
 
Everything seems to be OK, another Twilight Zone moment.

I'll agree, all is very strange. I'll admit I looked briefly at the semiconductor assembly instructions on George's page, but dismissed any ideas of insulating requirements. It looked like U1 and D1 are the only parts that require electrical insulation. The CCS (U2-U3) and the mosfets (Q1-Q2) shouldn't. Oh well.

My heat sink is not the standard one from the parts list, it is a solid CU sink that I had to build because DigiKey was out of the called for heat sink

Aluminum anodize is a pretty effective electrical insulator. I wonder if that has anything to do with it? I'm glad you have it working. Hope the gremlins don't return. :)
 
Last edited:
Another Twilight Zone moment. These Voodoo Electronics moments are really getting me down. I have always used the Vishay MKP 1837 as a bypass for any coupling cap. The Tublab SE still was not as stable as I though it should be, so just on a wim I removed the 1837's. This seemed to help quite a bit, the amp is quite stable now. The question is why would this .01uf bypass cap cause instability in the circuit.

Also, does anyone know of a better quality pot (5 turns or so) that will fit on the Tubelab SE board. I believe that some of the instability is resulting from these inexpensive pots and a 5 turn pot would make adjustment alot easier.

Lee Hankins
 
They could cause problems if they are leaky (DCR) or become leaky as they get warm inside the chassis.

These cermet pots are actually not intended for repeated use (take a look at the data sheet). They are intended for being set once and then forgetting about them. That said, they seem to work fine for most folks. There is a carbon comp pot that will drop-in. It's the same type used on Pete Millett's DCPP. I don't know of a multi-turn drop-in, but any of the larger ones can be made to work with a little creative lead-bending. Take a look through the datasheets on Mouser.