Rebuilding a pair of Frazier 106C's
These arrived here some time ago being doorstops while I finished some other projects. As I told Scott, these were going to be a side project. I finally started with playing them. First order of the day was to make sure they ran at all. And this they did. Following a piece of advice from the internet, I cranked the bias down to allow the tubes to run cooler. Frazier is running these tubes near max plate disappation at idle.
I ran some performance tests, but I could not get either amp to test beyond 40W/rms without popping their internal circuit breakers. Since these had unusual output impedances (much more on that later) of 5.4 and 12.6 ohms (as well as others) I tested the output trannys individually to see what was really happening. These output trannys have 6 of 8.4 ohm windings (not taps). More by accident than design, I discovered one of the output trannys had an open secondary winding. When I opened the tranny to tke a look, one of the output wires was shorted to the case. Or so I thought. Digging just a little deeper, I could see that one winding had never been connected at the factory. I opened the other tranny just to compare, and it was fine. I repaired the connection, but still never got either amp to push past 40W without tripping the breaker even though I swapped the trannys to each chassis just as a cross check.
During the power checks I measured the THD at 1KHz (at about 25W or so) which was low. I didn't make a note of it, but it was well below 1% and both amps tested the same. No unusual residuals on either amp.
After this they went back to door stop duties. Eventually they ended up in the living room for a listening test. I used a pair of re-built Advents. These are larger cabinets than original, loaded with original, but second generation parts and a custom crossover. The amps sounded good at first and we rocked around the house a few days getting used to them as you always have to do with a new amp experience. But after a while I grew weary of them. They were harsh and gritty sounding. Not in your face bad, just a "wear-me-down" thing that happened over time.
Well, Scott wanted them re-done, so let's dig into this design and see what can be made of it. When I had the trannys out, I measured the impedances involved and concluded they are running 12K plate to plate (p-p) at the design impedances. This seemed reasonable for a high voltage (full 400V plate) low current type power amp. Not much different than other TV output tube designs or some well known 807 units. The design was grid bias with a small cathode resistor to measure the bias. No balancing controls were provided, and the bias was set with one control for all 4 tubes. Screen voltage (150V approx.) was provided for using a 0A2 and a dropping resistor. Each screen had its own stopper. Basic power supply was CLC using a low Z choke from a bridge rectifier.
The original amplifier used a cascode input circuit based on a 12AU7 for the gain. This was feeding a long tail splitter also using a 12AU7 via a coupling cap. The design was well thought out and cleanly executed using a heavy duty circuit board to hold the component parts, while the tubes where mounted to the chassis. I like this better than PC mounted tubes when it comes to vintage amps. I traced every wire and resistor on the schematic and calculated what they were doing in the circuit. My impressions were good about their concept. Obviously they spent no small time getting this circuit right. There is a special feedback winding which returns its signal to the cathode of the lower stage of the cascode circuit.
I tried various mods to try to discover why this amp didn't sound as good as I thought it could. One of the things I tried was lowering the plate voltage in the output section by means of adding a 5U4 and a 5V tranny to the circuit. I was thinking of lower the load line position a bit and adding a slower warm-up as a bonus. This worked and softened the sound somewhat, but this alone was not going to save the day.
I set the amps aside, moving on to other projects. This happened quite often with these during all this process as I spoent no small time just contemplating the design and options.
I decided to work (rework) the load line. I had 6 output windings that could be combined in various ways and a target load impedance of 4 ohms in the Maggies. I also wanted to provide for an 8 ohm load just in case. When I re-worked the load lines and numbers with these reflected loads, I ended up with an output tranny of 8K-9K (p-p). We could be fine with that based on a load line using cathode bias. I estimated that about 70W could be achieved.
After re-working one of the amps in this configuration and setting the output windings according to my needs, I spent some time listening and gathering impressions. One amp only. I felt I was definitely on the right track. It still didn't sound quite right to my ear, but the harshness was gone. The bass loosened up a bit, which was a disappointment but not unexpected. Especially since I had reduced the global feedback about 6 db. I felt I was on the right track, but just missing the mark.
Back to doorstop duty while I mused on the situation.
I thought long and hard and even dreamed these amps for some time. I didn't want to change the basic input/driver stage setup as I felt they had done a great job. But the issue had to be drawn: was it the circuit design or the output trannys? These were meant for PA use after all, so perhaps the OPT's weren't so hot. I mean, generally trannys used in this service were pretty good, and in a lot of cases the same ones used for hi-fi, but these particular ones were unknown to me.
Well obviously, I'm not changing the trannys, so I'll have to see what can be done about the circuit. This is not the first amp I've seen using a cascode input stage, but it is only the second. Not a common circuit within hi-fi (noisy). I saw it in this design as a cost cutting measure against gain as opposed to using say as 12AX7. And if they did use a 12AX7, then they only would need only one triode from it, thereby wasting a triode. So the design chosen made sense, although they could have opted for one of the single triode tubes.
Another thing that bothered be just a little was the use of a coupling cap from the input stage to the longtail. They really didn't need to do it that way, but part of the issue was getting the input 12AU7 bias point to set into a good position. As I said, the circuit was well thought out.
I decided I would incorporate direct coupling from the input to the splitter. But first, I redesigned the front end as a SRPP stage. I needed more gain since I was giving up the tremendous cascode gain (even tho I planned on reducing the feedback), so I opted for a 12AT7 (later a 5965). I recalculated the power supply drops so my SRPP stage center point would be fit for direct coupling to the longtail.
The 12AU7 as a longtail splitter was a fine choice. Still, it has its work cut out for it, driving the paralleled miller capacitance of two output tubes per leg. I opted to try the 6CG7 in this position. The circuit was only slightly changed as needed for the new direct coupling. The total current through this stage is now nearly double the original.
So now here we are. The SRPP was seeing a good and constant load driving the longtail with its grounded grid at t2, just what an SRPP likes, and I had optimized it for that service. Global feedback was reduced partly to make up for the gain change and partly because I felt it was a player in the sound issue.
The output stage still carried its change, operating in cathode self bias operation. I tried a self balancing bias arrangement, but the loss of power was too great, cutting this huge amp to a mere 20W or so. So I settled for a 370V net plate at about 42mA per tube. Net plate to plate was about 9K.
Operating as above, the amp hit 50W before tripping its breaker (they were always tripping their own breakers). THD at 40W (now sustainable) was well below 1% at 1KHz. I still had only one amp set up this way, but I put it into service listening to it for some time. It sounded great. (Later without the breaker the amps makes 70W at clipping.)
Before and during all this, the power supply itself was re-worked. I left the basic screen regulation in place, changing only the dropping resistor and adding a bypassed zener to increase the screen voltage now that I was using cathode bias. I replaced all the power supply cap cans with new high-performace units which are now mounted under the chassis. I only slightly increased the overall capacitance. The huge (HUGE) power tranny's dynamics under load indicated plenty of depth, and I wanted to keep the recharging current low and the cycle fast.
A few pieces were cut from aluminum to cover the unused chassis holes. After rounding and smoothing these pieces they were bolted on and the entire chassis was then powder coated in satin black. The tranny bell ends were removed, blasted and powder coated a gorgeous metal flake silver. The cores were lightly painted and the choke outer frame to match.
Special holes were cut for the speaker connectors and the EIC power connector. The original circuit breakers were exchanged for conventional fusing.
Stainless steel hardware was used throughout except where not appropiate or unavailable in the sizes needed. I did however, use chrome acorn nuts on the trannys to spruce them up a bit and add some glitter.
The Transaudio's are a 6" two-way using a variation of the Emit tweeter. I've always liked these, although the tend to sound a bit heavy in the bass and lacking in the mids. However, I had replaced the original woofers with units that were not so heavy-coned, and that helped smooth out the response and brought the mids to normal. These small cabinet speakers are naturally very inefficient, which is why I had them tucked away.
I connected the amps to the Transaudio's first. Now I had something I could crank up! Blew me away! These amps are rock solid right on the money for these speakers, and I suspect therefore will do excellently with the target Maggies. It was a real pleasure to listen at the edge of the room's abilities without worrying about running out of power or headroom. Imaging was smack on, and the amps presented a sound stage of surprising depth. Deborah and I spent two days listening to these little speakers perched on top of the Advents, and enjoying the sound. Never missed the very low bass loss at all. Detail was excellent.
This experience has inspired me to renew my interest in building some larger amps. I have built a few pushpull amps in the past, and have always enjoyed them. It is now time for me to use some of the classic iron I've been saving these last 5+ years.
Of course, we eventually had to hook up the Advents. These are much more efficient than the Transaudio's. They also get a LOT deeper in the bass. So it took a while to aclimate to the new sound and to find the right positioning of the Advents for our room. But once settled in, the sound was fantastic - although not as extended in the high end as the Transaudio as expected. Imaging is a bit duller on these, although the sense of depth and localization was just as good. SO we spent an evening rockin' the house down with Stevie Ray, the Eagles, some Queen. Later I played some jazz so I could get a sense of how well the double-bass was represented, and it was clear as a bell and not at all muddy.
All in all these amps convinced me I needed a pair - or at least something in this scale - for my own use. I even thought about un-converting my Dyna ST70 just to remind myself of what that amp sounded like original, but I'll pass on that one. I'm now thinking more along the lines of an 807/1625 PP amp in the 40W range. POWER matters! As far as I am concerned these sound better than a Marantz model 8. It's not even a contest. Could the 6GT5 really be that good?
abraxasaudio(at)att(dot)net
