Jump to content

Eico HF-87 Clone Build


Curious_George

Recommended Posts

Mullard 5-20 circuit's always sound good to me. One of my favorite simple push pull amps that just perform well without any fuss. Of course different tubes but I have heard a lot of different tubes with this circuit and it's the circuit that sounds good. What feedback resistor value are you using and how much does it give you? Just a quick guess I think 4.7k should give you near 20db.

 

 

I'd keep an eye on the power supply with such a low value choke, you may need to increase capacitance a little more. The resonant frequency is high at 19Hz

Link to comment
Share on other sites

On 9/5/2022 at 2:29 PM, Curious_George said:

I left the feedback values (Rfb/Cfb) off for now until I dial them in. I'll do up to 20dB, that should be fine, no more. 

 

I used that choke (0.5H/300mA/30 Ohms) because I have a pair and the ripple reduction is quite significant with those capacitance values, but I will keep an eye out for instability. I could use a resistor like the original HF-87, but I wanted less initial ripple going to the power stage. 

 

This is a distant cousin of the 5-20 topology as you can see. 

 

8 minutes ago, Curious_George said:

The power supply appears to be ok with the chosen values. No issues (yet). I posted some comments above about the build and sound quality.

 

That's good, you were just creeping up into the audio frequency spectrum with the resonant frequency. If you do run into any issues just increase the capacitance if you want to keep using the same chokes.

 

-20db of feedback should be plenty to get great results. I add the lag network first then the lead network. The lead network is the cap across Rfb, it breaks upwards and will flatten out the response after the lag network (typically in the plate circuit of the first stage) is added.

 

The quickest way I have found to accurately find the amplifier breakpoints is with a 45° Lissajou figure on the scope. With it in X-Y mode place the scope leads to amplifier input and output to measure phase difference. Increase frequency until you get the desired rise/run 45° ellipse pattern on the screen, write down the first breakpoint frequency. Keep doing this until you find all the amplifiers breakpoints by 45° intervals. Once you know where the breakpoints are we then know gain margin and phase margin of the open loop amplifier circuit. I can explain this better if you want but I have found most people just sort of make a guess at comp networks and adjust until square wave response is better. I had the hardest time doing it that way and hated it so I wanted a more definitive way of knowing the exact phase response of the amplifier and to calculate out the feedback networks to get me about perfect right out of the gate. Sometimes I do  go back and move the lag network up if feedback didn't add enough bandwidth or down if the results didn't give me the stability I wanted and there is enough frequency bandwidth to do so.

Link to comment
Share on other sites

On 9/18/2022 at 2:31 PM, captainbeefheart said:

 

 

That's good, you were just creeping up into the audio frequency spectrum with the resonant frequency. If you do run into any issues just increase the capacitance if you want to keep using the same chokes.

 

-20db of feedback should be plenty to get great results. I add the lag network first then the lead network. The lead network is the cap across Rfb, it breaks upwards and will flatten out the response after the lag network (typically in the plate circuit of the first stage) is added.

 

The quickest way I have found to accurately find the amplifier breakpoints is with a 45° Lissajou figure on the scope. With it in X-Y mode place the scope leads to amplifier input and output to measure phase difference. Increase frequency until you get the desired rise/run 45° ellipse pattern on the screen, write down the first breakpoint frequency. Keep doing this until you find all the amplifiers breakpoints by 45° intervals. Once you know where the breakpoints are we then know gain margin and phase margin of the open loop amplifier circuit. I can explain this better if you want but I have found most people just sort of make a guess at comp networks and adjust until square wave response is better. I had the hardest time doing it that way and hated it so I wanted a more definitive way of knowing the exact phase response of the amplifier and to calculate out the feedback networks to get me about perfect right out of the gate. Sometimes I do  go back and move the lag network up if feedback didn't add enough bandwidth or down if the results didn't give me the stability I wanted and there is enough frequency bandwidth to do so.

.

  • Like 1
Link to comment
Share on other sites

4 hours ago, Curious_George said:

I just "guessed" at the compensation cap based on a quick frequency response plot. I will connect the scope and dial it in to see where it really is. This particular design does not have a "lag" network, only lead. The Williamson had both lead & lag. 

 

You can apply a lag network to any amplifier circuit regardless if it never had one to begin with, more times than not it will greatly improve stability. I have added them to lots of vintage amps including Eico. Some circuits will have an RC network or just a cap from plate to ground, it does the same job as plate to power supply node since the node will have a low impedance to ground anyway through the filter cap.

 

Same with adding a zobel network at the output, if an amp didn't have one and the specific circuit will benefit from one I'll add it in a restoration/repair. It can be on either side of the transformer but I typically use the secondary side because capacitance is cheap. In the old days it was probably cheaper to use a small value ceramic cap on the plate side instead of a higher value paper/polymer cap on the secondary.

 

On many vintage amplifiers these networks were omitted as the amp was "good enough" for most easy loads. When something comes my way I don't just try and put the circuit exactly to the schematic, I improve upon the design to make the amplifier better. Many of my friends or anyone that's handed me an amplifier greatly appreciates the extra care to improve upon things and often times it's an audible improvement.

Link to comment
Share on other sites

I was going to say I could look through my notebooks and try and find the Eico values for the networks but since yours is a clone with different output transformers it would be useless.

 

I can walk you through making the bode plot. It's actually pretty easy once you get use to what to see on the scope for an ellipse pattern.

 

 

it's one of the next plateau type things in amplifier designs. After I got the hang of it feedback is no longer a trouble maker and the designs will yield excellent square wave results.

Link to comment
Share on other sites

I have not looked at the square wave response yet either. I wanted to get it up and running and burn it in to ensure it was working ok. It seems to be fine and now comes the fine tuning. Thanks for all your suggestions, I’ll definitely post results of the network results and further testing. 
 

Maybe later this week I can do the Bode plot.

Link to comment
Share on other sites

11 minutes ago, Curious_George said:

I have not looked at the square wave response yet either. I wanted to get it up and running and burn it in to ensure it was working ok. It seems to be fine and now comes the fine tuning. Thanks for all your suggestions, I’ll definitely post results of the network results and further testing. 
 

Maybe later this week I can do the Bode plot.

 

Yup, the natural order of progression where fine tuning is at the end. I was discussing the highly cloned Western Electric circuits like the WE91A, naturally there is substantial high frequency roll-off starting around 10kHz which is why they added positive feedback to increase the high frequencies. Practically all clones will be unstable if you just copy the schematic verbatim. I have actually heard one Japanese builder claim that the amps should have major ringing and stability issues as it's part of the sound. I think the engineers at WE were not striving for that outcome, they worked the values to fit their specific amplifier with specific components and specific layout.

 

Similarly with the later 40's-60's push pull designs and global feedback implementation. Many builders just use the values given on the schematic which typically won't hurt anything, they are just not optimal and can certainly be improved upon with drastic improvements many times.  Just like the Mullard 5-20 or Eico circuit, the common cathode gain stage, long tailed pair inverter, and push pull finals typically UL, with 20db of feedback they perform very well. I have gone as high as slightly over 30db of feedback with the original Pentode input stage but the amplifier is just too sensitive and much harder to get excellent square waves. Most of us now just go with a triode for the common cathode input stage and a medium mu LTP inverter. Still gives excellent open loop gain for plenty of sensitivity and still allow for 20db or more of feedback. It's a tried and true design that gets built a lot to this day because it just works like glue.

 

It's one of the great tube amplifier circuits! s

Link to comment
Share on other sites

Although (it seems) a lot of people think the "old iron" was better such as Acrosound, Stancor, Partridge, UTC, etc than Edcor or Hammond, I have had good luck with new modern equivalents of those transformers. The main problem with buying those older transformers, is you do not know the history of the device. Certainly the old cloth covering is brittle and will crack as soon as you move it and/or you don't know if the core is OK, meaning if it ever had any heat problems to due a tube failure or oscillations, etc. Going with a new OPT makes life easier, but the compensation network will be different.

 

I'd like to hear an amp with 20% UL taps, I think it might sound a bit better than 40% taps. The 20% tap will move operation a bit more toward "pentode" mode and give you slightly more power too. According to the Philips tube book, 20% UL taps is optimum. Going to 40% makes less distortion but not by much. I doubt it would be audible. I think the 40% taps was made popular by Hafler after he published his work on UL back in the day.

 

With 20 ~ 40% UL taps, the output tube is also more forgiving of its primary impedance and it can vary from about 6k ~ 9k and performance does not change hardly any. 

 

The HF-87 circuit can also use 6L6's with a cathode resistor change. I might try it soon, just to hear the difference. Right now, input sensitivity is more than I like, but it is quiet, so I can't complain. It is right on the verge of being at the threshold of too sensitive because even though it is quiet, it is not as quiet as my Williamson clone which requires 1.65VRMS to reach full power. 0.550mVRMS will drive the HF-87 clone to full power and that is using a 6SL7 instead of the original tube, a 12AX7!

 

 

 

 

 

Link to comment
Share on other sites

3 hours ago, Curious_George said:

Although (it seems) a lot of people think the "old iron" was better such as Acrosound, Stancor, Partridge, UTC, etc than Edcor or Hammond, I have had good luck with new modern equivalents of those transformers. The main problem with buying those older transformers, is you do not know the history of the device. Certainly the old cloth covering is brittle and will crack as soon as you move it and/or you don't know if the core is OK, meaning if it ever had any heat problems to due a tube failure or oscillations, etc. Going with a new OPT makes life easier, but the compensation network will be different.

 

I'd like to hear an amp with 20% UL taps, I think it might sound a bit better than 40% taps. The 20% tap will move operation a bit more toward "pentode" mode and give you slightly more power too. According to the Philips tube book, 20% UL taps is optimum. Going to 40% makes less distortion but not by much. I doubt it would be audible. I think the 40% taps was made popular by Hafler after he published his work on UL back in the day.

 

With 20 ~ 40% UL taps, the output tube is also more forgiving of its primary impedance and it can vary from about 6k ~ 9k and performance does not change hardly any. 

 

The HF-87 circuit can also use 6L6's with a cathode resistor change. I might try it soon, just to hear the difference. Right now, input sensitivity is more than I like, but it is quiet, so I can't complain. It is right on the verge of being at the threshold of too sensitive because even though it is quiet, it is not as quiet as my Williamson clone which requires 1.65VRMS to reach full power. 0.550mVRMS will drive the HF-87 clone to full power and that is using a 6SL7 instead of the original tube, a 12AX7!

 

 

 

 

 

 

Yes the original uses a 12AX7 which will also effect compensation changes and the sensitivity is certainly too high at 380mV to full power. Many of these old amplifiers were meant to be driven from a crystal pickup directly so they were shooting for a high input voltage sensitivity compared to our modern amps.  I Increased the amount of feedback on many of the ones that came in to over 20db, added the lag comp network on the plate of the 12AX7 and changed the lead network cap value.

 

Many of us builders that clone these circuits, well not verbatim just the topology because they have too much open loop gain and don't really like more than 30db of feedback so we change tube types to lower gain, especially if the circuit had a pentode like the hf35 or the original Mullard 5-20. I'll usually triode wire a 6AU6A for the first stage and use a 12AY7 for the phase splitter. I have used all 12AU7's, 12AT7's, EF86 in triode mode for the first stage and many more. I was partial to the triode wired pentode input and the 12AY7 LTP, it was one of the better combinations I tried.

 

The issue with UL when sharing the same winding as the plate is when the plate swings down near ground with 43% taps especially the screen gets pulled down with it and makes the tube weak in this region of operation. Of course this happens at higher output levels and doesn't show very well on regular bench testing a sine wave against an constant load. Normal dips in impedance at bass frequencies down near half the original impedance the power tube can't supply enough current in that low screen power of operation when the plate pulls it down near ground. Look at a curve tracer when the screen is at only 50v, all the curves get squished together and compressed, then when the load halves the load line goes much more vertical on the current axis. Drawing it up shows how non optimal the load line is against the squished curves. With a nice stiff regulated screen supply in pentode mode this doesn't happen. So in the real world with real loads a well designed pentode amp often will outperform an UL amp in the higher power region.

 

One neat experiment that showed how negative feedback is one of the major reasons for performance in transistor amps is making a +/- 300v supply and using two 12AX7 tubes in an opamp type configuration. When setting the feedback resistor to achieve a gain of 40db distortion wasn't measurable from my equipment with near 150v peak to peak output. The other advantage of transistors is complimentary pairs which tubes just cannot do ever, we will always only have NPN type devices with tubes. I always wanted to try that 12AX7 opamp circuit as a front end stage to drive a big single ended triode output stage that requires lots of swing. But it never happened due to it being kinda ridiculous. But if we can remove the driver distortion from these types of amps then deal with local feedback around the output triode then we could achieve very low distortion levels for the entire amplifier. Oh well something to play around with in the future.

Link to comment
Share on other sites

I can see how UL mode would cause compromised operation near max power. However, there were millions of units made (kits and DIY) with UL as the chosen topology. Even the Marantz 8B was UL. 

 

The book "An Approach To Audio Amplifier Design" by G.E.C (England) has a bunch of amps, some fairly high powered. Good reading.  

Link to comment
Share on other sites

Cappey - I ran a 20kHz square wave through the HF-87C (c = clone) and it looks fairly clean, minimal ringing. Looks like the same response as pictured in a lot of the advertisements when they were selling these kits back in the day. I'm still going to dive deeper, but just giving you an update. 

 

More listening today. It's hard to tell but, I think this amp definitely sounds better than the Williamson I built. The Williamson clone was the first tube amp I built... nothing can sound better than it! 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...