Fixing a Chinese Tube Amp

Being a semiconductor man I still couldn't resist tackling a high vacuum glow transistor project. This Chinese built tube amplifier, made by Shenzhen Valve Audio Lab was imported by a Dutch outfit who also modded it at the owner's request. Unfortunately both the Chinese and the dutchman didn't really know what they were doing so I set out to put things right.

The design is a clone of the famous Dynaco ST70 kit. Unfortunately the clone has some defects. First of all, the input tube, the totally unknown 6F2 (which turned out to be an ECF82) is intended for use in TVs as RF mixing tube. The triode is meant to be used as oscillator and the penthode as mixer. The Dynaco's 7199 is an audio tube, optimized for low level operation. I did not pursue the consequences of this but I did notice some hum problems during testing. The Dynaco design uses a weird g2 configuration that causes the screen grid voltage to drop to a very low value, somewhere between 10-20 volts. I don't understand how it works but the Chinese duly copied it.

The bias circuit was totally fucked up. When adjusting a tube the three other tubes varied with it. So I fixed it somewhat to the original Dynaco spec. Now not only the setting were stable, adjusting was easier as the range was not cramped at the operating point of the output tubes. The input capacitor will not blow up because it is not running at its rated voltage anymore.

Now the design was taken care of I went to the original request which was to move the ugly input jacks to the back. As the chassis is steel this was not easy but eventually I got the phono jacks liberated from a satellite receiver mounted where I wanted them. They were wired with the wonderful low-noise Phonoflex cable. I also fixed the volume control which was badly abused (broken axle tab). Really should be changed for a nice Alps unit. Anyhoo, time for testing!

The Chinese tubes turned out to be unmatched. The right channel had two completely different grid bias points for the rated idling current (30 mA), 47 and 53 volts. The other channel was more or less equal at 41 volts. Fortunately my junk box had a large number of original Philips EL34s and I managed to get two matched pairs. All tubes are now around 44 V.

Time for testing. Tube amps have a notoriously high distortion but this one was a bit extreme. This turned out to be another fault in the original design: the anode and cathode resistors of the phase splitter weren not equal. Dynaco specifically requires matched resistors here but the Shenzhen people decided to make them different and even on my scope the distortion was apparent. But when I made them equal it was still considerable. Only when swapping the resistors distortion dropped to 1.5% @ 10 watts. I used the same trick for the other channel. The left channel was even worse: 6.6% with the Chinese tubes which dropped to 1.6% after the swap. Changing the Audio Voice (Audio who?) tubes to Philips did not make much difference but I kept them because "MINIWATT" looks übercool...

Now I was ready for an A/B listening test. I concocted a switch box with banana sockets to switch the outputs of the VAA and my own trusty Philips DFA888 to my speakers. I used the tape rec output to drive the VAA from the Philips. First I calibrated both amps with a 1 kHz test tone and then I put on my favorite CD. What I got was a lot of noise in the right channel. What the fuck?!? It turned out that there was another design flaw in the VAA! The input pentode was drawing grid current, a sure-shot way of fucking up your sound! A quick look at the Dynaco schematic revealed why: the Chinese designers left out a cathode resistor which is necessary to properly bias the 6F2. So I started to experiment with a resistor inserted in this location and found that with the grid current gone the anode voltage became too high. A quick fix for that was to raise the screen grid voltage a bit with another resistor. This improved things considerably, distortion at 10 watts dropped to 0.7%. At the onset of clipping (50 W) it was still only 2%, down from 4%.

Finally I could do my test. Wired it all up and hit play. The tube amp sounded decidedly different from my transistor amp. The latter had a cleaner bass and a more detailed high. The bass was because my speaker has a lower impedance which the output transformer apparently can't cope with. The highs I dunno, could also be the output transformer. Maybe I should do some more testing, even though it seemed that power bandwidth was OK, only the distortion was pretty significant at 10 kHz.

 


full_view_orig

vol_control

input_kludge

serial_shenzen

nu_input_rear

nu_input

phonoflex

nu_vol1

nu_vol2

now_what_is_it

resistor_swap

nu_bias_c

schematic

bias

el34_closeup

final_test

 

The story continues...

Copyleft 20 May 2012

Back to the homepage