View Single Post
  #5   Report Post  
Posted to rec.audio.tubes
Patrick Turner Patrick Turner is offline
external usenet poster
 
Posts: 3,964
Default Tales of a Heathkit W7-M amplifier

On Sunday, 30 September 2012 08:32:57 UTC+10, Analogdino wrote:
Hi, Vacuumlanders, I'm seeking advice and opinion on a vintage Heathkit audio amplifier – the W7-M. I was given one a few weeks ago by an old gent who built it back in the day. It was a bit grubby but reasonably well constructed... although one resistor (6AN8 plate) was the wrong value for the last 50 years! Now fixed. Of course, I am compelled to refurbish it to add to the “toys” for use once in a while... This is an odd duck... it uses s/s Si diode doubler B+ rectifiers from a 180 VAC PT secondary and a 1/2 wave Se fixed-bias rectifier from a 40 VAC secondary. Tubes are two EL34's and one 6AN8 (pentode amplifier, triode phase splitter.) It has a physically large OPT, actually larger than the PT, and tests about 3500 ohms impedance plate-to-plate. The schematic is at http://www.vintage-radio.info/heathkit/index.htm To keep this short (various unproductive farting about omitted!) ... B+ caps replaced (two were dead, two reformed OK); fixed bias filter caps reformed OK; all paper coupling/decoupling caps replaced; EL34's very weak (pity!), so replaced with 6L6's to hand; 6AN8 tested good; DC voltages look about right. Variac-powered up carefully... no smoke! B+ good, fixed-bias voltage good, AC power draw OK, signal gets to speaker. So far, so good.... But, max power output is lousy! The W7-M is listed as a “55 watt amplifier” but the most I can get at 200 Hz is a bit over 10 watts. At that level, one half of the waveform “soft clips”. More forensics done (with NFB off)... it clips asymmetrically in the 6AN8. With the o/p tube grids driven directly with a P-P source (a centre-tapped audio transformer, primary driven by a 10 watt s/s utility amplifier), I can just about get 25 watts into an 8 ohm load at clipping – but at least it's symmetrical, hard clipping, i.e. evenly maxed out 6L6's. Confirms poor 6AN8 performance, so that needs fixing in due course. Here's some test data from the o/p stage-only test using new, Russian-made 6L6's, all just at clipping, fixed bias balance and level optimized: 115 VAC in: B+ at max load = 464 VDC Across 8 ohms: 38 VAC, peak to peak, 200 Hz, for 22.5 watts RMS 120 VAC in: B+ at max load = 478 VDC Across 8 ohms: 40 VAC, peak to peak, 200 Hz, for 25 watts RMS Bias around -30 VDC per tube... not much effect on the above, left at max. Changed Se rectifier for 200 volt s/s Si diode, gives a couple of volts more available bias. Repeated tests: 115 VAC in: B+ at max load = 457 VDC Across 8 ohms: 40 VAC, peak to peak, 200 Hz, for 25 watts RMS 120 VAC in: B+ at max load = 481 VDC Across 8 ohms: 42 VAC, peak to peak, 200 Hz, for 27.5 watts RMS. Just a smidgeon better, but way off 55 watts! Older, used but good 6L6's give results just a bit below the above, but very close. So, where to from here... some of my printable thoughts to date: 1. W7-M is not even a modest collectible... so, dump it “for parts or repair”, no big loss to the world! After all, it's only a old PA amplifier. 2. The W7-M is as scarce as hen's teeth, a priceless collectible, so keep it “as is” for posterity... Ha, ha, had to say that! 3. W7-M was never “55 watts” - that was pure hype, so settle for 25 watts with a rebuilt 6AN8 front end. 4. Ditto, but re-design and rebuild the front end as a true Williamson, i.e.. double triode amplifier/phase splitter and double triode driver stage (there's space for an extra tube base hole in the chassis.) Fix the expected VLF instability (addressed elsewhere) as OPT, although large, is certainly not up to Williamson specs (plate-to-plate primary inductance likely nowhere near 100H – that costs money!) 5. Put it on the shelf “as is” and wait for inspiration... at this point, an attractive option! 6. Any other ideas? Just thought... was Heathkit playing the IHF (?) “peak power” game. IIRC, there was a promotional hype back then to advertize amplifiers at around twice the true RMS power rating to sell basically crap equipment. But I don't think HK did this. All replies gratefully received, recognizing, of course, that none of this is life changing. After all, it's a hobby... but I hate unfixed stuff! Cheers, Roger


You should find that at some low value RLa-a you will get 55Watts with fixed bias abd class AB1. The PO amounts you quote would be high % class A1 but do the load line analysis and all would be revealed. Maybe the amp will have enough heater power for 6550/KT88/KT90 and then you can get what you get now but Ea will swing lower with low load, but if you want hi-fi, just keep Ia and Ea so Pda at idle = 25Watts approx, and then you can settle for the 22W in nearly all pure class A1 - even if you connect the 6550/KT88/KT90 in triode.

The Heathkit input driver stage looks like an RCA or Dynaco schematic with bean counter minimalist use of two tubes, pentode input, triode CPI. The pentode could generate lots more gain if its anode load is bootstrapped off the cathode load of triode CPI, see the RCA typical designs used. As it is, if the amp had about 15dB GNFB, then it'd be fairly insensitive because the pentode has anode load of a low 33k,and if gm = 0.002, A = 66 approx, so for say the max 24Vrms at anode, you need approx 0.37V Va-k, and if there was 20dB GNFB then Vin = 3.4Vrms approx. The pentode input tube needs to make slightly more anode signal voltage than applied to each EL34 grid. The added gain using bootstrapping results in a lower amount of GNF, ie, ß is lower, for the same amount of GNFB, because open loop gain without GNFB mght be 4 times greater.

Patrick Turner.


Read my website for many more ideas about how to improve the amp using LTP driver stage with CCS for cathode, and triode input, and for all info on how to stabilise the amp unconditionally, ie, make it LF stable with say 20dB GNFB even with no load at all connected, and stable with any value of pure C across output terminals, between say 0.01uF and 3uF.

www.turneraudio.com.au