Page 2 of 2

Re: Big Video Amp

PostPosted: Thu Jul 13, 2017 6:09 pm
by Doug Coulter
Step by step, inch by inch.
I changed the power supply for the opamps to be + 12 and -22 volts so I could go a little more negative on the output, and maybe run some screen voltage.
To accomodate that and keep the nice (for this) property of it all being cutoff with no input, I changed the offsetting resistors from 1k to 619 ohms. I also put a 100k resistor to the negative rail on the positive inputs (which are all tied together) so with nothing plugged in, it's at the absolute negative limit output of about -20.1 volts to the control grids. With the input shorted to ground, it goes up to -19.1v or so. Both are in cutoff...with +15v on the screens (so...depending, maybe I'll increase that voltage).
For now, not, as I may want to use kickback in the output circuit to get higher peak output voltages at some point - so complete cutoff gives me more of a switch than a linear amplifier, and with the arbitrary waveform generator I can adjust the offset to what I want anyway, should I want to run "more or less pure linear amp" (knowing full well that this isn't going to be the king of fidelity no matter what).
6kd6's are rated at 7kv peak plate volts. I've run them past 12 kv peaks in flyback with no issues if they're turned fully off. They make great switches and decent amplifiers.

During this testing, I had one "rogue" tube in there that didn't match the others, and the plate current in cutoff went to 3-4 ma. When I put in all matched Zenith tubes, it went down to the numbers on the schiz. (I prefer Sylvania for power tubes, so if anyone has a stash, let me know). I have some super-weird GE tubes that are two small sweep tubes in the same envelope in parallel....GE, don't get me started, I bet they're almost entirely unlike a real 6kd6, despite markings.

But, so far so good. I have an APAP machine blowing air into the chassis underside over the opamp heatsink I made, then out and up past the tubes...nice quiet blower for this (at one time we scored a bunch of these machines, reprogrammed them to their max output, and use them for cooling other fusor things too - nice and quiet as these things go).

I'm using one opamp per two tubes here as the specs say I can and it seems to work so far (I haven't tried AC real signal inputs yet, just DC bias tests). When trying to drive the control grids positive, they limit to around 5v instead of the 10v or so the opamp puts out in theory. The 47 +11 ohm resistor (effective) plus the opamp current limit agree with this.

Those extra 22 ohm resistors in each tube grid were just there (so I didn't change it, they were mechanically convenient) and used to be all that was there. I added the 47 ohm in series from the opamp output on the advice of the data sheet for driving big capacitive loads - a 6kd6 is supposed to be around 40pf input capacity, and here we have two - a pretty big load for this preamp. That 619 ohm offset resistor is a little more conductance than the 638 ohm calculated, but in tests it works right - the opamp input isn't a perfect virtual ground.

It's hard to believe nothing oscillates here (yet - now go back to your room, Murphy)...I did "the right things" but it usually also takes some luck when there's this much gain and this much bandwidth at the same time in the same box...

preamp schiz as it is now

Re: Big Video Amp - smoke test

PostPosted: Fri Jul 14, 2017 4:06 pm
by Doug Coulter
No smoke!
It does seem to have a very slow turnoff time, which is puzzling to me...
This is with a 2100 ohm load (2k wirewound + 100 ohm wirewound), tapped off at 100 ohm with a .047 uf cap to a 10k to ground and the scope. So, roughly 20::1 attenuation.
Doesn't seem like the capacity of the cable to the plates (about 4' of coax) would make it that slow...I'll have to test more.
This is not full drive and you can see the DC bias is below effective cutoff here.
scope, input vs output of amp
AmpTest.png (5.04 KiB) Viewed 456 times

So, doesn't turn on at zero - this is easy to adjust for. 300+ma at 1.5v input...not bad. ~~ 630v plate swing here.
I'll have to measure that cable capacity - a really rough slide rule calc shows me ~ 200pf which seems high, but then it was a HV-connector cable, not a normal type of coax (scrounged thing, I didn't think a BNC type connector would handle some of the possible HV peaks here).

Edit: Facepalm, surprised I don't have bruises there. :roll:
Cable + tubes: 220 pf
Cable = 88 pf
Approx 22 pf/tube, then.
Data sheet tube output capacity: 16pf

So, theory == practice, which does in fact "blow my skirt up" (how often does that happen?). I'll assume the rest of the capacity is due to the spider setup that interconnects all the plate caps - it's pretty leggy so <6pf/leg plus the wire to the connector seems pretty reasonable here. Something I'll have to think about later on when I care about fidelity of waveform.

On the other hand, if I reset the bias or more likely increase the screen volts so it comes on right as the input comes off zero, I'll have so much possible current per tube I might be able to take a few out...heh. And people wonder why this isn't a one day job.

Re: Big Video Amp

PostPosted: Fri Jul 14, 2017 5:21 pm
by Donovan Ready
Doug, do you still even have a slide rule? Mine is a K&E log log decitrig that's gone yellow with age. Sort of like me... :mrgreen:

Re: Big Video Amp

PostPosted: Fri Jul 14, 2017 7:07 pm
by Doug Coulter
Yes, I have my own along with the ones I inherited from my Dad. In this case, I used a "special" from Shure, handed out free in WWII to the engineers who had to actually do stuff. It's super handy, and just solves the reactance/resonance equations - directly, no brains required (or very few - two drunk neurons are enough).

Here's the artwork. The original was in cardboard, and with not much work you can print these, cut things up, glue them to cardboard from the back of note pads, make a couple narrow spacers, use some stables, and bingo. If you do anything with reactance from tuned circuits to filters...this saves a ton of time.

You can do this on the cheap if you just use thick printer paper...but this is worth getting right (for me). Glad we got one scanned when it was only a few decades old. Family treasure for a family of engineers.

Heck, I even have a special "employee award" rule made from invar and it's a couple feet long. Collector's item. Kinda hard to shove in the nerd holster and carry, though.

I don't use the GP slide rules except to scare the kids. I do use this one all the time...just try writing a program that gives you the abilities this does this easily (solve for any variable by just sliding to match it on one of the scales, then reading off the others). I've thought about how for a few decades go, sometimes this is just the best.

Re: Big Video Amp - more screen volts

PostPosted: Sat Jul 15, 2017 2:43 pm
by Doug Coulter
I put the screens at 27.9v or so, which was the handiest voltage I had around that was in range.
With that, the idle current with input open is 4ma, with input shorted to ground it's 7ma (which is about 1v less negative on the grids in this setup).

With a 1.6v input signal, offset .8v, it swings the plates nearly to ground (would be the next major div). Full drive would be around 3v input.
This is with the 2.1k plate load. Note the scope point has an additional ~20::1 attenuation before the probe. Every major div is 200v for the yellow trace.
The board doesn't like my scaling to 640x480, so raw here
Screen28v.png (3.49 KiB) Viewed 431 times

A little cleanup, mount the fan better...and this will be done for now.

Re: Big Video Amp

PostPosted: Sun Jul 16, 2017 11:24 am
by Doug Coulter
Almost dressed up - no problem with someplace to go.
the nasty underbelly

topsides (it is the Navy's iron after all)

Gotta mount that fan...I think. But there's some cooling for those opamps and the underside generally via an APAP machine that makes 15 cm Hg pressure - a lot goes through that hose.
cooling for the semiconductors and bleed out to tubes

Now just need to make a nicer line cord and put it in place. I also need to finalize the drive software for the arb signal generator and so on.
Hopefully Bob and the tubes will forgive me for just current limiting the grid inputs, and using a 3.5 amp slo blow fuse for now...

Re: Big Video Amp

PostPosted: Tue Jul 18, 2017 10:14 pm
by Bob Reite
Hopefully Bob and the tubes will forgive me for just current limiting the grid inputs, and using a 3.5 amp slo blow fuse for now...

me laughs Considering what you are driving it with, grid current limiting should be fine. You'll build a remote resettable overload relay for the plate supply the first time you pop a fuse in the middle of a run and have to shut down the whole system so that you can go out to the shack and change it.

Re: Big Video Amp

PostPosted: Wed Jul 19, 2017 11:48 am
by Doug Coulter
Yeah, probably I will. I wouldn't have to shut down the whole world, but would the signal sources and power on a couple things (have that going on now). No need to mess with the vacuum stuff for example.

One issue that's not put to bed yet is how many "channels" I get to over there, effectively. I can add more stuff wirelessly, but in a high EMI world, that may not get the required reliability. To reset a relay, sure, as if that's needed there wouldn't be noise at the time, but other things....maybe not so much. I have one fiber line to there, and the converter is inside the main control raspberry pi box, and drives that pi. If I were to put in a switch back at the bench 10' away or so to drive more stuff (and then would need more wire<>fiber converters), I'd have to trust the switch never ends. That pi is about used up, and every time I add another hole for a new signal into that box...well, you know about having to test it against everything to ensure that the additional EMI getting in didn't break something that used to work.

The inrush on that plate supply is pretty fierce too. I should put in an inrush limiter. It's acting like that mechanical switch isn't going to live forever.
The original had a bunch of 5ar4's in parallel where I now have the RF tubes...this has 3 amp silicon diodes...

The tube warm up time has advantages in my bench testing. I can watch the meters and so on as they warm up, and if something's going to be in trouble, yank power before they get all the way there...handy.

Right now, my remote switched 240v over there for the main HV doesn't have its own neutral, or I'd run this from that...the best thing might be to add that wire, painful as it might be. That one is SSRs and control is buried telco wire back to my op position panel - I feel like that's rock solid (SSRSs are rated 600v/40a each, have MOVs inside, and are on heatsinks).