120w valve amplifier.

[Refugee]

I have seen sweet corn cobs ground up and used in a vibrating tray to polish metal parts especially aluminum.

[Diabolical Artificer]

Thanks Brian, I didn't think to straighten the wire that way, good tip. I've been doing it by running it though a tightly held rag.

A bit more done but a few mistakes made. The amp chassis is 7 1/2" by 12 1/2" and sits between two end pieces like the cross bar on a H. I mounted this 2" above the amp case floor, but this meant that the OPT hit the amp case top cross bar, so therefore wouldn't be able to slide in. So had to lower it.

Lowering it to 1 and 13/16" maent that the CCS heatsink was too "high" ( see pic 1) and would hit the bottom of the case. So off with the old HS but I didn't have one suitable so I had to cut one off an old car stereo HS that I've had in the junk box for years but was too weird to be used. This meant bodging the CCS board a bit, but not too badly.

I've also hit a potential snag with the chassis power connector (see pic 2) which came off the old Tek scope. It's a McMurdo red series RSD16, which ITW still make. I wasn't going to butcher the Tek plugin so had to buy the male connector. These ain't cheap at £21.83 each, but I found some in Oz at £4.50 each so got 6 which makes the postage cost ok.

However reading the datasheet - see attached - it says contacts = 5A "All contacts (simultaneously)" which may mean each pin/contact is rated at 312mA each. This means that I'll need to double up the pins for the HT and ground and use eight (or would it?) for one 18.9v 1.5A heater supply. Each pin is 19 thou thick by 1/8" wide. The McMurdo octal socket pins are 15 thou thick by 3/16" roughly and I presume they were designed with the idea that valves like the 5U4G which has a htr current of 3A, could be used. looking at similar/suitable type connector's, the Amphenol blue ribbon connector which is similar has a pin rating of 5A. Looking at the Tek plug ins I have two that use 15 small valves like the 12AT7/6AU6. At a rough estimate that would be 15 x 300mA = 4.5A. As far as I can see Tek used only two pins for the htr wiring loom, so maybe I have it wrong and am talking bo*****s. Feedback most welcome.

The amp chassis bolts will be counter sunk and the front plate will have a thin plate on top, as per Tek scope and plug in.

That aside here's some more pics....

Andy.

[Diabolical Artificer]

I've more or less finished putting the amp chassis together so gave it a test. I powered it up without the 12BH7 in the socket, but with the wiring and components present , see attached schematic.

The front end tested ok but the ECC82 phase splitter read -3v on the cathodes! Typical, it worked ok in it's rough state but not neatly put together. After much taking apart and head scratching and swapping the pot on the CCS for a 1k preset (was 300 ohm), I stuck the 12BH7 in and the cathode voltage went up to 10v, bang on.

I'm pretty sure I know why this is, but thought I'd see if any of you chaps would like to have a go at working out why.

Andy.

[Nick]

Anode loading. Without the 12BH7, the ECC82 anode load includes the 1K grid stopper and 12K cathode resistor as the diode is forward biased. This pulls down the anode voltage and thus the cathode too.

With the 12BH7 in situ, the diode is reverse biased and we're back to just grid current.

[Diabolical Artificer]

Beat me to it Nick, just replied to your PM,. Yep, your right, couldn't put it better. Sorry no fruitcake and a fiver out the till for the winner. : )

Andy.

[Diabolical Artificer]

I've nearly finished fettling the chassis or rack unit to accept the two amp "modules" and have temporarily wired in the PSU. I managed to fit it all in though this has taken a lot of measuring, jiggling, trial and error. I've had to make several compromises because I didn't really think everything through. To be fair to myself though unless I'd actually built this beforehand, I couldn't have thought about everything before hand. Still, if I'd made the amp module 1/2" smaller on the length and width, life would have been easier.

Starting with the PSU, the mains enters the rack unit through an IEC kettle doodah, goes into a big 32mm/1 1/4" NOS fuse holder and is switched both L and N with a NTE thermistor on the big toroid tfmr primary. I tried fitting caps across the SW to surpress arc, but this means the PSU isn't properly off. The thermistor should extend SW life somewhat anyway.

Without the thermistor even with a soft start circuit, there is a hell of an inrush current. At first I used 100 ohm resistor's on each capacitor "bank" , but this didn't reduce the current surge or delay the rise of HT voltage. I'd obviously miscalculated. I found some big vitreous enamel resistors in the shed, that came of the Tek 454A scope that donated the ceramic wafers I used on the amp. So I used two these, a 700 ish ohm on the "bottom" 225v HT PSU, and a 1k5 on the top. this delays full HT by around 8 seconds, these R's are then shorted by a relay, two contacts for the top HT, one for the bottom. See schematic.

The relay delay circuit is powered by a little tfmr, GR etc, bottom left, pic one. You can just see it under the cap board. the delay circuit is on the PCB on the left. A TIP122 Darlington switches on the relay. It's overkill for this, but works well. I tried using a 2SD880 at one point, but it oscillated.

Next, middle on the PCB is the bias/CCS PSU. I had to change values of the first RC filter resistor. There were two 1k5 100u RC filters, but the first dropped the OP voltage too much. I'd also put a 80v zener on the OP but had to take it off. There is a bias adjust pot on the amp, this dropped too much voltage. I experimented with the "fools bias resistor" ( a R to stop bias from being set to 0v) to stop too much voltage being dropped. 1M and a 50k pot gives -16v to -40v and doesn't drop too much voltage.

Last, on the right of the PCB is the 12.6v PSU circuit that powers the htrs of the triode gain stage and PS. I screwed up here and wound the tfmr secondary that powers it with too high a voltage, EG 34v when rectified and smoothed. this means the LM317 got very hot indeed. So I had to retro fit a dropping resistor of around 20 ohm's (two big cement R's on right) to give 19v IP to the LM317. Not happy with this bodge but it would mean re-winding the big toroid . No way jose! : )

I keep writting PCB, but it was made using a dremmel and fine diamond tip to cut the traces. As you see it and the relay is on a hinged panel to make access easier. This hinged panel took about a week to get right!

The fuseboard needs re-doing. I used veroboard but am not happy with the result. It has gaffa tape on the chassis to stop any shorts at present which on the finished amp will need something better. Also there are no covers on the HT fuseholders - an oversight because as the bridge rectifiers have caps across each diode to suppress switching noise, HT is present on the fuse holder.

Lastly for now, as you'll see in the last pic is a panel meter. I managed to get two the same off ebay for £20. They are the same NOS military 50uA but have different scale plates. I'm thinking of using these to make PPM/VU meters. I saw there are cheap Chinese VU meters that come with a driver board, but thought I could do better.

That's about it for now, TFL, Andy.

[Diabolical Artificer]

Here's the pics. Technical difficulties this end - slow broadband.

A.

[Diabolical Artificer]

I tested everything yesterday and all worked well for a while. Previous to that I'd noticed the cathode follower resistors made up of matched 10k and 2k2 5w wire wound, were getting very hot, I could smell scorching. I measured the current and found they were running at 20mA, which is a bit too much. I'd also noticed that the cathode followers wern't biased quite right. They should have been at -10v but were at -3v.

I replaced the 2k2 R's with 5k6 which reduced dissipation to 15mA, the resistors still get warm, but not overly so. After some thought I realised I could set the bias with the adjustable CCS of the phase splitter. This meant moving the bias point of the ECC82 triodes a bit, but found with the ECC82 PS set to -6.6v ( they were running at -11v, center biased, but not running in the linear region), the cathode followers were now set to -10v.

So as mentioned at the start all was going swimmingly, I was testing and setting the bias of the OP stage. This is set by a dual ganged linear 10k pot which balances the two sides of the OP stage. I noticed that it wasn't working as well as it did on the "prototype", and after a while prior to switching off I unplugged the lead from the sig gen. A loud screech issued from somewhere, I hit the off Sw.

I noticed I had nasty oscillation on the OP and after some investigation I found the screech was coming from the OPT. I tested the OPT for shorts and whilst doing so I found the tag of the pin on the valve base that the screen tap of the OPT connects to, was broken. This accounts for the difficulty of balancing the OP stage. These are decent NOS Cinch octal bases so not rubbish, but I did think when i got them that the tags looked a bit thin, and I had to get off some scale off each tag with a file prior to soldering. This must have put undo pressure on the tag. I started to unsolder every tag, then realised I had a spare base, so nicked a pin/tag off it to replace the broken one.

The only problem with doing a good job of soldering leads to valve base tags, IE making a good mechanical joint by wrapping a few turns and cinching with pliers, is that it's a bugger to unsolder without wrecking the hookup wire. Anyhoo, did that, screech still there, OPT not shorted. So I took out the input, PS and cathode follower, screech gone.

So something is broken/amiss with one of these three stages. Today if i get time I'll see if I can get to the bottom of the problem. At least the OPT seems undamaged, thank the gods. It seems odd that disconnecting the sig gen should cause such a catostrophic failure. I've unplugged/turned off the sig gen gen loads of times before, without issue. However I was using a different sig gen yesterday.

Two steps forward, one back. this is a bit like playing snakes and ladders, A.

[Nick]

Ferrite beads on the 807 grids?

[Diabolical Artificer]

It was working fine Nick. I'm not sure it's oscillation as such, the scope trace is weird, havn't seen the like before.

Whilst doing the textbook taking of voltages I leaned my arm on the chassis which flexed it a bit, fault disappeared. So there's a loose connection there somewhere I guess. will investigate tomorrow, been out at Spalding radio rally.

A.