Where are the 833 amps?

[rcavictim]

Originally posted by James D.


I love the Tesla coil pics - but they scare me! One of my jobs as a junior engineer with Decca was to help test their ceramic insulators. I had to stand in a huge Faraday cage with the insulator under test connected to a 500KW transmitter and then spray water over the insulator to cause rf arcing!!! Then the insulator was examined for damage. Great when it worked like that but every now and then the arc would come back down the water stream towards you... It would normally earth around you as one had rubber kit on and used a plastic water gun but it is dammed scary to have a mini lighting strike going on around you!!!


ciao

James

James,

I did some work for RacallDecca here in Canada some years back with broadcast antenna insulator testing. I got a chance to sit in on some testing in the US where they rented time on a decomissioned submarine ELF transmitting station converted to do HV RF testing. There was a half million watt xmitter there at ~30 KHz. It fed a HV generating resonating coil the size of a school bus wound with Litz wire that was the size of a broom stick handle. When they powered it up, if there was a ladbug or a fly anywhere on the HV buss it would launch a flare and they would have to shut down and then re-energize. Great electronic bug zapper! I am familiar with the water spray behaviour you spoke of (Mil spec water!). Interesting field, pun intended.

Sorry for the off topic.

[Brian Beck]

I hesitate to put this monster before you, because it is so insane, and not finalized in final form yet. But I’ve described it elsewhere in DIYAudio, so I’ll be brief here. Several years ago I built a direct-drive amplifier for electrostatic loudspeakers (ESL) using 833s. These tubes directly connect to the ESL stators without transformers or coupling caps. The 833s drive Quad ESL-63s happily, bypassing both any OPT but also the step-up transformers in the Quads. However I went all out on this amp because I needed a test source for a new ESL design project I’m involved in. It was also a test bed for several design ideas. It’s totally insane, impractical, unsafe and very hot, of course. Each channel uses four 833s in a SRPP bridge running off +/- 4000 supplies. Output is 4800 V rms, or 6800 V zero-peak.

I will post pictures of the breadboard version only for now. The final configuration is a stereo pair housed in half-height rack chassis, one chassis per channel. A glass enclosure will keep fingers away from high voltages, and act as a plenum for forced air cooling, required due to eight 833s in the room dissipating 2000W. A dedicated A/C supply duct and a return duct will be routed, so that heat will not directly impact the listening room, a necessity here in Florida. However, home renovation, followed by two hurricanes, and now a move has conspired to halt final configuration progress.

Here is one channel's worth of tubes cooking.

[Brian Beck]

This picture shows the breadboard of one channel. The input circuits are up front; the 833s are farther back. Behind the 833s, barely visible and leaning at an angle for servicing is the +/- 4000 volt supply. The 100 lb. transformer for the HV is from a Collins T-368 transmitter. It's bigger than a car battery. I was lucky to snare a pair of these beasts.

[Brian Beck]

Here's the input stage breadboard, all DC coupled with level shifters. I've made liberal use of 6SN7s

[Brian Beck]

I used the cores from salvaged 300VA variacs as my filament transformers. I kept the primary intact and wound 15KV-rated silicone insulated wire around the core to achieve 10VAC at 10A for each tube. The cathodes swing at high voltages so the high voltage wire was necessary. This picture shows one such transformer connected to a test socket. The final eight filament transformers are sandwiched in a elastomeric clamping arrangment to keep the secondary windings from buzzing around.

[Brian Beck]

And, finally, here's a blurry scope shot showing music playing at the output on the Quads at robust, but not crazy loud, levels. Each division is 1000 volts (I developed a special compensated HV scope probe for this).

[SY]

That is seriously frightening.

What are the 9-pin tubes?

[rcavictim]

Brian,

You are one hip dude! My kind of DIY'er. Hats off and all that. I hope you have current limiting between the amp and the electrostatic speakers to prevent possible explosions and fire.

I'm planning quite an amp at the moment using a pair of 450TH's in push pull making as much as 2 kW audio power but as far as pushing the envelope, it won't even come close to how radical this thing you have conceived is. Nice! Hope it doesn't hurt anyone, ever.

[rcavictim]

I am renewedly examining the possibilities of making a very high powered monoblock for primary application as a subwoofer amp as mentioned above Tube candidates are a pair of Eimac 450TH's in push pull. Key to this is of course getting a suitable hunk of iron that will work as an output xfmer. I wasn't sure until today but I think I have that covered. Yipee!

I have a huge Hammond plate xfmer with heavy cast end bells and a house project number so I cannot look up specs. It has 1710-0-1710 HV secondary and primary with 0-100-105-110-115-120 volt taps. Turns out 120 volts into an 8 ohm load is about 2 kW so that is right in the ballpark of the output that this amp will have. Turns ratio makles 8 ohm load appear as ~7 k ohms plate to plate which couldn't get much better if I ordered a real o/p xfmer. DCR is only 60 ohms across the entire HV winding. Primary L is high at 34.6 Henries. That is what I need for good bass response. There is a lot of iron here. It weighs about 100 lbs. It came from a WW-II xmitter. I have been packing this thing with me since high school. Bloody glad I did!

I just finished doing frequency response testing by putting the 600 ohm o/p of my HP audio oscillator through a 6.8k R to the full HV winding and loading the 120 volt secondary with 4 and 8 ohm R's and watching the response on the o-scope. Ruler flat from 40 Hz to 8 KHz. Response at 15 Hz only down about 1 dB. It looks like with NFB this will make full spectrum. 15+ KHz at the high end. I replaced the generator with a CD player and the o-scope with an audio amp and speakers. At microwatt power levels the transformer sounds pretty good on full range. I had to boost the treble on the monitor amp but that brought the highs back.

Just for curiosity I grabbed a similar but smaller (700 VA) Hammond plate xfmer, same secondary voltage, somewhat newer and it only had 25 Hy's primary L. I wouldn't be surprised if my huge one was wound for 25 Hz. That makes it a rare chunk-o-iron indeed and better suited for my application.

As far as a HV PS capable of delivering enough for a pair of 450TH's making 1.5 to 2 kW output I am all set. I have a complete HV power supply I got from a hamfest sale of some of Fred Hammond's personal ham effects. It contains a big 2000-0-2000 volt plate xfmr rated at 2.25 kVA. I forget the make...oh yes Hammond. Also has SS diodes good for the job, a suitable huge choke and filter cap bank and bleeder R. Including 450TH tubes and their odd 7.5 volt, 20 amp filament xfmr's (which I grabbed with the tubes out of some xmitters I stripped) I have all the expensive parts to make one only of these wacky monoblock amplifiers. Might as well!

I'll begin by breadboarding the output stage and PSU then do tests driving the beast with a variac and split winding power xfmr for P-P grid drive at 60Hz. If this looks good I'll design a xfmer coupled driver, P-P of course, later.

Now I'm far from being an expert in these matters of tube amps and output xfmers and build stuff very successfully most of the time based a lot on instinct. My hands seem to be pretty smart. The bench breadboard will tell me ultimately if this is a go or no-go idea but if there is something about this reapplicated plate xfmr that I'm not seeing feel free to step in and comment.

[Brian Donaldson]

Brian

That's pne crazy amp. You've done the name proud.