G9 problems continued...

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Bjorn Zetterlund

Well-known member
Joined
Jun 3, 2004
Messages
123
Location
London, UK
Hello all!

My previous post was very rudely interrupted, so I thought I'd post an update.

My G9 problem (an 8ch point to point version) was that the cathode voltage for the first stage was far too high. Alas, due to recent copyright disputes, the copyright of the value of this voltage no longer belongs to me, and so I'm not at liberty to disclose this information. Let's just say it was somewhere between 90V and 100V, and actually closer to 100V. My HT supply was a little high as well, something like 295 instead of 245, so this would have contributed some. However, I did try one channel with a 230V HT supply, but the cathode voltage was still about 90, so I obviously have something wrong here. My next step will be to wire up one valve out of circuit with just the HT and bias resistors, and see if I can get that to settle at the correct operating point.

What turns out to be the second problem, is the fact that my HT supply does not appear to be regulating; as in the tl783 does not actually drop any voltage. I have tried replacing the regulator, and the surrounding circuitry measures fine. I have not tested the three zener protection diodes, nor the diode that is wired in parallel to these. I just thought I would run this by you folks to see if anyone spots anything wrong in my redesign for this supply to power 8 channels instead of 2...

The 783 can supply up to 700mA, and I thought this must be plenty for 8 channels. It is mounted on a big-ass heatsink, before you ask! Therefore, I decided to use the original G9 supply, but with suitable scaling of components. The input series resistor (R33) I changed to 120 Ohms, so as to drop the same voltage but at four times the current of the original supply. C14, the resevoir cap, is three 330UF in parallel. The output filter (R37, C15) is actually located on each channel rather than in the (external) PSU unit. Given that each is supplying only one channel rather than the two channels of the original G9, I decreased R37 to 220 Ohms to drop the same voltage for half the current, and I made C15 47UF instead of 100UF. That all seemed like common sense to me, but I thought I'd double check with you all, just in case I've done something stupid (and I obviously have done somewhere down the line!)

I have tried this supply both with the 8ch G9, and also with a dummy load of about 3K, but the results are the same. Any thoughts on this?

:sam: :sam: :sam:

Bjorn

...are there any UK techies that know if you can get Sam Adams here??
 
Bjorn,

I would try removing the zeners and the protection diode for now. The TL783 should not suffer as you don't have a huge inductive load on your regulator. The zeners appear like overkill to me, but that is just my opinion.

Also, make absolutely sure that the TL783 is hooked up right:
Pin 3: Input
Pin 2: Output
Pin 1: Adjustment node

Tamas
 
Also, the TL783 won't start regulating untill you draw some current - that is, your tubes has heated up!

If you want regulation to set in from the start, you need to shunt it with a resistor dissipating some 5-10mA.

Jakob E.
 
[quote author="tk@halmi"]I would try removing the zeners and the protection diode for now. The TL783 should not suffer as you don't have a huge inductive load on your regulator. The zeners appear like overkill to me, but that is just my opinion.[/quote]

Hi Tamas,

Please Don't try this!

The Zeners are there to keep the input/output differential voltage within the safe specs of 120V. Whenever you turn on power with empty secondary reservoir capacitors, the in/out differential will be much larger than this untill they're charged. This can easily kill the 783 - I've suffered this several times.

The 1N4003 protection diode is needed as well, as it protects the 783 from reverse differential voltages lower than 0.6V. We need this when powering down - when primary reservoir capacitors gets lower voltage than what may be stored in the secondary reservoir capacitors!

I know there's often overkill protection on regulation circuits - but these are actually needed. remember that we use this device far out of it's intended SOA.

Jakob E.
 
Bjorn,

You mentioned that the PSU is external- how long is the lead you're using to connect the PSU to the Octo-G9? I've had problems before with regulators not regulating, and upon 'scoping the output, it's appeared that the reg has gone into an HF oscillation. You could try placing a cap at the output pin of the TL783 to ground- a high voltage (500V) 0.1u poly may do the trick. I just wondered that it may not like driving a long cable directly out of the reg without a decoupling cap.

Just a longshot!

And yes, we need to see high-res pictures to be able to help further :wink: :green: :thumb:

Mark
 
[quote author="gyraf"]

Hi Tamas,

Please Don't try this!

Jakob E.[/quote]

Sorry, my bad. The TL783 is rated at 125V max indeed.
Would lifting the ground of the regulator with zeners work?

Tamas
 
Hello again and thanks for the replies!

I just ordered some more parts from RS today so as to do some further testing. Unfortunately, they have discontinued the TL783, I haven't checked yet if this is just RS or if it has indeed gone out of production.

All that I have to add at the moment is that the regulator is connected properly, and that I have always tested the psu with a load; both the actual G9 as well as a dummy load (there's about a 100k or so between my thumb and my pinky :green: ). No really, 3k or something.

At the end of the week I'm going to first just try my two TL783's out in more forgiving low voltage circuits to verify that they actually work, and then double check everything, for the 10th time!

As far as potential oscillation goes, yes it has a 3m PSU cable...and there is also a small inductor at the G9 end (just to kill RF...not for any smoothing as such). I'm pretty sure I also have a reasonably hefty electrolytic on the PSU after the regulator, but I can't swear to that at the moment. From what I know, that should be enough to swamp any naughtyness, no?

Anyhow, thanks for chiming in everybody, and yes I will certainly do pix and a little show and tell site...that is once the bitch is actually up and running :wink:

Thanks!

Bjorn
 
......"I just ordered some more parts from RS today so as to do some further testing. Unfortunately, they have discontinued the TL783, I haven't checked yet if this is just RS or if it has indeed gone out of production. "


That sweird because I just got a TL783CKC part no. 168-7282 from RS on the first of June. They didn`t have it at the trade counter & had to order it in for me !!!
 
Hi Rob,

On the TI website they are listed as active. The two 783's that I ordered from RS for this PSU wasn't more than a month ago...who knows, maybe it was just a temporary screw up.

Bjorn

...just checked the RS site:

TL783KC has been discontinued (RS 302-255)

TL783CKC is still available (RS 168-7282)

I must have gone for the KC version, just odd that they didn't offer a replacement when I asked? Oh well, I shall go for the CKC version tomorrow.
 
Right...some further experiments!

I haven't done any work on the PSU yet, so that puzzle still remains.

What I did tonite, was to connect up one ECC82 with the DC conditions of the first two valve stages of the G9. That means I included R6,7,8,9 for V1A and R23,24,25 for V1B. For this test I used a variable HT PSU from another device. Having spent a long time trying to find some fault with one of my G9 channels, I wanted to do something really simple to investigate where I was going wrong. My measurements follow below. Note that HT in this case is the voltage after the 1k/22u filter between the two 'tubes' of the G9; 'node' is the junction of R6,7,8. The voltage at V1A cathode is connected to a 63V cap, and on my G9 has a voltage of about 100V, which is how I knew something was wrong in the first place. Each voltage has three values; the first is for a National ECC82 from RS, and the second two are chinese origin audiophile grade- purchased from billington export and meant to be tested and of very good quality.

V1A

Plate: 160,158,157
Grid: 47,48,50
Cathode: 77,80,81

Node: 74,76,78
HT: 234,234,234


V1B

Plate: 83,78,75
Grid: 0,0,0
Cathode: 2.2, 2.3, 2.3


Having established that the results seemed to have a reasonable spread between different valves, I thought I'd try lowering the HT until the V1A cathode voltage dropped to 60, which would be acceptable for C7. I found that I had to drop the HT to 180 for this. According to the G9 schem, that would mean (245-180=) 65 volts dropped by R31, which is 1k...hence it must pass 65mA. But I just read in one of Jakobs posts to another G9 question that each G9 channel needs about 8mA, so something must be wrong here! I do use 6.3V heating, so it will be slightly hotter that the standard G9, but surely it can't be due to that? And should the 'node' voltage and the grid of V1A differ by about 28V? That would suggest a grid current of 27uA?

So just where am I going wrong with this :roll:

Bjorn

...and these DC readings do correspond to the values of one of my G9 channels posted over at RO.
 
Your cathode circuit is messed up.
You are showing anout 30 volts between the grid and cathode, and without having built this thing yet, I can tell you that you have a problem.

I hate to mix high voltage with solid state, something always pops.
I would junk that regulator circuit and just use a si mple RC pi filter.

Sometiimes you get gremlins on the high tension sec. that are way above the B+ voltage, only you won't know it.
It snaeks in and out before you see it, just long enough to nuke the reg.
Also, regs can be noisy!
That would be an abosolute sin to have a noisy regulator feeding your first stage.
Comapany's don't check regs for noise!
So many times, noise regs slip past qc.

Luck of the draw.

I can see why Jakob would want a nice tight regulated voltage , especially with european power flucuations, so not bad mouthing it, he has to make these projects universally sound, to work on all continents, so ...

Measure your resistance from grid to cathode (pwr off!) and I'll bet you a T4 that you will find your problem.
exception:the tube is bad which I highly doubt, as you have more than one channel doing the same thing.

cj
 
Thanks CJ!

So do you mean I get 8 T4's if all 8 measure fine? :grin: :grin:

Will do that next. Could shit sockets cause this kind of trouble? I didn't use anything fancy; just whatever is the RS standard one- some kind of white material.

Bjorn
 
Yes, 8 T4's.
You can sell em on evilbay!\


Sockets won't cause that radical a problem.
The only problems I ever had with sockets is on power tubes, arcing and that sort of thing.

Oh crap. Could ultra sonic osc. be causing that?
I have seen it do weird things to voltages.

scope it out at high freq.
 
Ok, will try a scope on the HT as well.

That would mean two PSU's with some kinda oscillation, though. Both with transistors, but one (that I used to test with) uses 'proper' high voltage transistors that have already been working ok to the best of my knowledge for two years or so. And with my original/proper G9 PSU, the regulator is not actually dropping any voltage...would that not mean that it inherently can't be introducing any kind of oscillation?

I can't help but think that one of these days I'm gonna have a real doh! moment...I think I must have just stared myself blind on this one. Just for fun, what would happen if you had the grid and cathode reversed on the first valve?

...and I am not worthy of another 8 T4's; not until I have given my first two a home, at the very least! In fact, if you're right, I'll send you a couple of sowter input trannies. The different types I used for my 8 channels of G9 were taken from an old Midas monitor desk. All are 1:7's, but slightly different types. If you win, you may do with them what you will!

Bjorn
 
You have too many variables

Check each one and mark it off your list.

the 783 is a good regulator. Should work
Find the datasheet and read the mininium regulating current.
disconnet the regulator output from the tube stuff.
Pull out ohms law and figure what resistor will give you that current at the voltage you have it set for. then reduce your resistor to ensure you are pulling greater than the mininium load. Make sure the resistor can handle the watts if not have fun with series / parallel strings of resistors.
Once that is done you need to make sure the regulator is regulating.
Use a voltmeter and a VARIAC change the input AC voltage
make a table of input and output voltages. Is is regulating.
then even if it is working get a scope use a X10 probe and make sure the output is not oscillating it could be high as 10 Mhz so watch out. Your probe
and scope should be able to handle the voltage if not throw them away.
You did put .1uf and 1uF caps on the output of the regulator directly at the output pins didn't you? look at the datasheet and follow the directions exactly. If you change caps there repeat the test above.

Make sure the reg is not toasted and the protection diode is in correctly.
It will bite you every time.

Now that was easy.
Leave the "load resistor" in place for the next tests.
Bring each B+ wire from each tube channel section out in the open. Use the PS with load resistor to check
each channel individually take notes of voltages and repair / replace tubes as needed to get every (channel) working as needed.

then remove the extra load resister and wire it back up.
Try 1/2 of all the channels then add each channel make sure all is working
as you add channels.

If you did the above right your unit nows works.
Not to be rude or sound condesinding you sounded like you were running around like a chicken with his head off. Check each section first before you
connect the next section. If it does not work remove sections to isolate the problem. Take good notes.
 
[quote author="gyraf"]some 15-20mA @245V HT, 600mA@12V Heater, and some 10mA Phantom.

It will supply more, if you need it...

Jakob E.[/quote]

Jakob, what's your theoretical max on the B+? No how to's required, I'm just trying to figure a $$$ budgeting query.

The tube manual puts the 12AU7 at 10.5mA at 250 plate volts (in Class A amp,) how does the G9 get away with 8-10mA per channel in the circuit?.
 
[quote author="bradzatitagain"]The tube manual puts the 12AU7 at 10.5mA at 250 plate volts[/quote]

That's not how tubes work. How much current a triode pulls depends on bias, not just on plate voltage. That spec must be some kind of recommended operating conditions that has nothing to do with the G9 circuit.

Each triode in the G9 probably pulls about 2 or 3mA... but that's just speculation on my part :wink:.

Peace,
Al.
 
Al, in the characteristics and typical operation section, w/ grid circuit resistance at 1M using cathode bias. But, okay, let me ask this another way.

I have four used power transformers with 12.3 and 6.3 (Xs 2) filament windings that used a 6X4 in the original circuit they got yanked out of. Tube manual states the DC output current of a 6X4 is 70mA, so, if the 8mA per channel figure quoted previously here (which I would guess would be max current draw) is accurate then each xt rectified w/6X4 ought to power 8 channels of G9. The only caveat would be unknown max filament current provided by the xt; it was heating 9 12A*7 tubes; 16 might be a stretch.
 

Latest posts

Back
Top