16V/6K8=2.35mA. Its not very normal for Q203 to lose 45% of its nominal at 10V VDS tested IDSS when in circuit under Vbe for VDS, rather a 20% loss is what they average. Is it a verified 5.22mA at 10V Q203? I knew of the GR0E ambiguous origin batch, with almost the right looks though, that had a bit higher Vgs(off) and bit softer curves that could lose more than 20% but not of other suspect house numbers. If Q204 Vbe is around 0.61V then the rest should be doing well. But I am not comfortable with the weird current ducking for Q203. I would replace it with a 200 Ohm resistor across D-S to compare drift.
P.S. Yes 38Vin is enough for 32Vout.
P.S. Yes 38Vin is enough for 32Vout.
There is also the 5K trimpot in the voltage reference, shouldn't I fact that in too?
I will try another 117 tomorrow, all done for the night.
That is to add up the final Vout max. But any drop across a series resistor in the loop tells the running current. 5K+6.8K=11.8K*2.35mA=27.73V+1.9+1.9+0.6=32.13V there you are it computes within a hair of experimental error even. 
See about the 200 Ohm resistor test that I also suggested. But measure Q204 Vbe first.
See about the 200 Ohm resistor test that I also suggested. But measure Q204 Vbe first.
So I replaced the Q203 and adjusted R203 accordingly and now everything is falling into the right parameters.
I have 16V across 4.67k R203, which gives me about 3.4mA, the value of Q203.
I used a different batch 2sk117 and it worked right away so I probably had a faulty one.
Now I'll let them burn in for a while and re check everything. Time to start ordering the cases for the preamp!!
I have 16V across 4.67k R203, which gives me about 3.4mA, the value of Q203.
I used a different batch 2sk117 and it worked right away so I probably had a faulty one.
Now I'll let them burn in for a while and re check everything. Time to start ordering the cases for the preamp!!
Full success finally 
I don't know what is going on with K117 faulty JFET samples percentage but I have also chanced on very few odd ones that fail right away without an apparent misuse reason or act strangely. What was the new one's full marking and typical IDSS measurement when with 9-10V across testing?
I don't know what is going on with K117 faulty JFET samples percentage but I have also chanced on very few odd ones that fail right away without an apparent misuse reason or act strangely. What was the new one's full marking and typical IDSS measurement when with 9-10V across testing?
On the positive it is around 0.60. The Q204 is probably a bit weaker in value, I didn't write the values down after measuring for populating the board, and will do so in the future for troubleshooting.
I let the regulators warm up and Vout drifted about .15V/.20 from cold to hot and remained unchanged after that. I had the regs powered on for about 45mins and checked 3 ou 4 times after hot.
One interesting thing is happening. I have this attenuator and control board powering the regulators up and down via the power relay on the transformer board.
Sometimes, when I touch the transistors legs or the Vout pads for measurements on the negative regulator the unit turns itself off or changed input channel. As if the multimeter is interacting with the negative regulator and sending an input to the control board. This happened about 50-60 percent of the time.
Everything is still wired in my table with no proper grounding and layout, and in the final assembly the power input board and raw supply will be in the PSU box, but it was a strange behavior nonetheless.
I let the regulators warm up and Vout drifted about .15V/.20 from cold to hot and remained unchanged after that. I had the regs powered on for about 45mins and checked 3 ou 4 times after hot.
One interesting thing is happening. I have this attenuator and control board powering the regulators up and down via the power relay on the transformer board.
Sometimes, when I touch the transistors legs or the Vout pads for measurements on the negative regulator the unit turns itself off or changed input channel. As if the multimeter is interacting with the negative regulator and sending an input to the control board. This happened about 50-60 percent of the time.
Everything is still wired in my table with no proper grounding and layout, and in the final assembly the power input board and raw supply will be in the PSU box, but it was a strange behavior nonetheless.
They both seem like working properly for readings and drift now. Within some tolerances. That touching thing probably provokes transient oscillation in the concept of the specific load and now wiring... Try the two wire mode also to see if it is more insensitive than the four wire mode in the final installation when cabling length and dressing will be optimized.
Finally got around to do some testing. This is for a BA3 FE balanced pre with two + and two- regs. in the case bottom 17in x13in. with around 110ma load, 360ma CCS and the temp of Q106 is only 50C.
Sounds very nice. I am sure the BIIB regs have played no small part! Thanks Salas. Intelligently designed boards, easy to work with and great help.
A couple of questions:
a) I hear some hiss in the speakers from the preamp. Not much but can be heard till about a foot from the speakers. I have used Wima MKS4 4.7u in all positions. You do mention in the guide about the tradeoff between the organic sound and some noise coming thru, so can/should this be reduced somewhat. I am only using a CD source, no phono.
b)Bit OT but I have read in the DCB1 thread you advising to set the DC offset at less than 5mv. At 5mv this would translate to 100mv at the speaker outputs of an amp with a typical voltage gain of 20. Is this much DC acceptable. I dont know so I ask.
nash
1. In a line preamp it could be due to the preamp hiss itself mainly, depending on its gain. If the PSRR is little and the gain is high some could be coming from the rails also though. How high gain you decided in yours? Are you using very sensitive speakers also, like big FR or horns? In dcb1, that there is no gain, many used film cap (Hypnotize version) with no hiss. But in your case you could try C101 & C201 100uF Nichicon FG or KZ instead of C102 & C202 film to make sure there is no added hiss from the power lines. If no hiss difference just keep the one solution you like better in SQ. Lytics can strengthen the bass and round the edges for instance while films be better on female voices.
2. 100mV final offset could still be safe for speakers but not elegant. I have serviced some brand name older amps with 40-120mV integral offset that played well for years. Depends on a specific amp's current bias a lot too. If the amplified offset messes more than 10% with it, it should be controlled to less offset. Most DCB1s are built down to 1-2mV but we had to draw a guideline max safe spec somewhere. That's for mating DC input power amps. With capacitor input models it does not matter of course. A DC detect and open power relay speaker protection system should be installed on any DC coupled power amp's output anyway. Especially when the full system chain is DC coupled. A bad source hooked up or a component failure somewhere along the chain can not be excluded in a system's lifetime. Then its bye bye woofers.
2. 100mV final offset could still be safe for speakers but not elegant. I have serviced some brand name older amps with 40-120mV integral offset that played well for years. Depends on a specific amp's current bias a lot too. If the amplified offset messes more than 10% with it, it should be controlled to less offset. Most DCB1s are built down to 1-2mV but we had to draw a guideline max safe spec somewhere. That's for mating DC input power amps. With capacitor input models it does not matter of course. A DC detect and open power relay speaker protection system should be installed on any DC coupled power amp's output anyway. Especially when the full system chain is DC coupled. A bad source hooked up or a component failure somewhere along the chain can not be excluded in a system's lifetime. Then its bye bye woofers.
My total gain is 10db but I am presently using SE in and out into mono F5TV3. I measured the AC signal in and the corresponding out and got a db gain of 2.5db. The hiss is there at pretty much the same level thruout even with the volume pot completely off. Speakers are vintage B&W DM6 with around 90db efficiency, mind you the hiss is not objectionable. More curious than anything else. There is no hiss with my passive autoformer VC. Would bumping up C102 film to 10uf from the present 4.7 help appreciably or perhaps just marginally?
Is the relay system used in the DCB1 a timed relay?
Thanks. nash
2.5x maybe? If dB seems low. In dcb1 its just a timed delay and fast release to avoid passing power up/down dc transients, yes. Detect and protect if adopted should always be done at the end just before the speakers. I don't think that 10uF can make an appreciable difference if the hiss is part PSU related though. You should confirm with an electrolytic. The rails must also be checked with the scope in case there is some oscillation or interference that translates to noise. If there is no scope maybe trying two wire mode can give an indication.