|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
96.11.127.78
I progressed with experimenting on my RH84 type circuit.
Hooked up an LM317 CCS as per the ver. 2 schematic on the EL84 cathode.
Voltage checks seemed to indicate correct function.
It worked for about 20 minutes, then the right channel dropped out slowly.
Too late last night for any troubleshooting, but the LM317 was connected directly to the socket minus any heatsinking.
Follow Ups:
These are the changes I made to RH84 ver. 2 schematic.Mostly based on the parts I had at the time and the fact that I was converting another circuit.
R1 = 400K , R2 = 220, R3 = 20K, R5 = 400K, R6 = 180
C1 = .1uF, , C2 = 33uF, C3 = 50uF
D1 = (2) 6.2V Motorola zener diodes in series
B+ = 302VDC T1 = 5K/8
Operating Points: all DC voltages
EL84: P = 293, g2 = 292 K = 10
12AT7 P = 124 K = 1.3
THD = 1.52% @ 1.17W RMS & 2.43% @ 2.75W RMS 3W RMS before clipping
THD measurements @ 1KHz with sine wave input
Summary: I should have build first on a breadboard to make measurements easier to take and followed the signal circuit schematic as per the design ver. 2 using a higher quality OPT than AES PT-31.
That being said, the subjective listening experience was more than good enough on 811B or a set of Magnavox 2-way horns.
I will not make advisements on the power supply configuration, but I did use a LOW DCR type - not low in uF.
Thanks to all for their contributions to this thread.
Edits: 01/24/15
The screen voltage on the EL84 seems off, if B+ is 300V with 2 x 6.2V Zener, it should be 283V... Of course, with the parts you had on hand, the operating condidtion for the 12AT7 also deviated from the rev.2 design values. But I am still surprised with the distortion level, it's nearly 3x higher than what the simulation from AK suggested. Oh, well if you like sound, it does not really matter.
I checked the part number on the zener - IN4733A 5.1V zener diode.
C2 302 - (5.1 X 2) = 291.8 @ g2. I measured 292V at g2.
Sorry for the confusion.
As for the THD measurements, the designer indicated that the OPT would be an influence.
The computer sims stating THD 1% @ 4W RMS represent an ideal/best case scenario circuit.
I did not find much real performance data from other builders on RH84 online, if you can find any links for us, please post them.
If I can improve on my current circuit, I will post the findings.
I would suggest a good set of doorbell transformers.
"I would suggest a good set of doorbell transformers."
My circuit does not have enough transfer efficiency to use a doorbell transformer as an OPT.
Most of the wiring is generic 600V stranded copper.
I only used the mil spec wire as signal ground runs.
Ah, that makes more sense. You don't find many measurement data on RH amps because very few people bother to test them, since they seem to like the sound. There was one gentleman that did, and like you he could not replicate the THD figure but I don't think it was so far off as yours were.
Since there is inter-stage distortion cancellation, the actual tubes used in the amp, make a big difference in the amp's performance and no amount of simulation can give you that data. But I am not sold on the idea that the OPT can make that much of a difference, especially since the tests were done at 1kHz, so even the cheapest OPT has a pretty good characteristic at that frequency. The results would be much worse had you tested the amp at the low and high frequency extremes.
In any case, you only need about a Watt with your efficient speakers, so it seems to do the job well.
I've looked at Heyboer SE 14W 25:1 vs. AES PT-31 .At 100 Hz , the Heyboer had 3X less distortion at 1W than the AES.
The bigger OPT have a definite advantage in the lower bandwidth Hz's.
With biamping, the little OPT's are good enough to use for the horns depending on where you are crossing over.
ARTA Steps can plot distortion vs. Hz and I have seen what happens to the distortion profile of a tube amp due to changes in biasing, OPT's, cathode bypass values, tube rolling , etc.
With the PC and a decent dual channel scope, you can find out a lot about what is happening with low power SE tube amp circuits.
Edits: 01/25/15
You would be amazed at the difference in distortion when switching from a small "barely enough" OPT to a "serious" OPT.
This is particularly evident as overall perceived output power and bandwidth.
If you are into simulation, play with the primary inductance and/or primary DCR.
******
http://rh-amps.blogspot.com/
Certainly the quality of the OPT can effect the bandwidth and distortion of the amplifier, but, at mid-frequencies, it should have barely any distortion, unless it is defective, so I don't think this alone would explain why the measured distortion is so much higher than what the simulation suggested.
Will be curious to hear your impression of adding the LM317 as well as your distortion measurements.
You rock DT!
1KHz sine input @ 1W power output :
THD = 1.27% THD+N = 1.29%
2nd harmonic = -38.16dB
3rd harmonic = -50.44dB
4th harmonic = -71.41dB
5th harmonic = -76.49dB
Voltage gain : 6.7
B+ off rectifier = 255VDC.
There is a wide variation in PS configurations by various builders.
Bandwidth -3dB was 17Hz and 18052Hz @ 1/4W.
OPT are AES PT-31 SE 25:1
IMO, the biggest improvements will probably be realized by the quality of the OPT used.
Thanks for sharing.
I was curious what kind of changes did you have to make to accommodate the lower B+?
It would be interesting to see the effect of putting in a SS rectifier (increased voltage) or a different tubed configuration, just to see what it does to your distortion results.
FWIW, I tried bypassing the 5V4 with SS diodes and that gave me a B+ of about 305VDC, close enough.
I changed the dropping resistor to the driver B+ cap to 10K ohms.
Put the LM317 on a tag strip away from the tube socket with additional heatsinking.
Unfortunately, the CCS was became unstable from initial power up and the B+ was constantly moving around.
Removed the SS diodes for lower B+ and the CCS worked properly again.
Didn't get into the troubleshooting process, but I think I should just get the exact parts values listed on the RH84 schematic if I have another go at this type of circuit.
The designer benchmarked 1% distortion at 4W RMS for this circuit and I did not come close to that level of performance.
Ah well you tried.
Let us know when you go at it again.
On the other hand, while you have the B+, why not try the first version of the RH84 without the zener diode and LM317?
I measured the operating temperature of the LM317 w/heatsink.
It was 85F, the extra B+ voltage did not seem to make much difference.
The voltage between out and adjust across the 27 ohm current set resistor was only .985V instead of 1.25V
Something is affecting the CCS operation of the LM317 when the B+ is at 300V.
I sleep on it and try a bit more tomorrow.
No need to sleep on it.
Bias voltage too low = cathode voltage too low.
In order to function properly, LM317 needs a voltage drop across it. This is not a low drop regulator. Drop voltage plus reference voltage equals cathode bias voltage.
Depending on Ug2, Ua, tube condition, etc. this might be too low for the LM317 and the best sign is reference voltage below spec (1.25V).
Probably your tubes are old or Ug2 is too low?
Did you remove the cap from g2 to ground?
Needless to say, all this affects performance. Spice sim is ideal, but you might have other issues in your test circuit like old worn out tubes, leaking caps, faulty zeners...
******
http://rh-amps.blogspot.com/
First, I removed the cap bypassing the CCS.The B+ and other voltages appeared to stabilize.
Measured 1.25V on the output tap of the LM317.
Removing the cap from g2 to ground did not have any effect on stability.
I will re-measure distortion/spectrum at 1W to see if improvements are realized and post findings.
Thank you for your assistance.
dt 677
Edits: 01/22/15
The cap from g2 to ground is superfluous and while it does not affect DC performance, it probably does affect AC performance (bandwidth).
The leaky or otherwise bad cap was probably influencing your output power vs distortion. Besides, at this power levels, 40-50V across the tube affect performance more than you might imagine.
Last but not least, 5k is not the same as 6k, particularly for a pentode.
******
http://rh-amps.blogspot.com/
i thought the lm317 needed to be bypassed with an SE circuit?
After removing the CCS bypass caps, I measured with ARTA at 1W and got horrid distortion , THD = 42%.
I didn't realize that the bypass cap was needed for correct circuit operation since I never tried LM317 as a CCS before.
I took out my scope and looked at the output of the amp, the resulting waveform was like a sine on the positive and a weird looking distorted square wave on the negative. It was really odd looking.
Next, I checked the driver stage and it was function normally.
After checking the EL84 plate, I saw the odd distorted signal again, so I knew it was probably the CCS.
But it was not until I read your post that I decided to bypass the CCS with some 22uF caps on jumper leads that everything worked again.
I'll have to try another set of 100uF bypass caps and see if anything changes.
Needless to say, not the most fun evening but it was informative, LOL.
Of course it does.
But the above is about DC performance improvement after elimination of leaky cap. The cap must be good. As for voltage value, twice the bias voltage I.e. cathode voltage is probably the best choice.
******
http://rh-amps.blogspot.com/
With B+ off the rectifier (UF4007 full wave) of 300VDC :
Total power output before clipping : 3W RMS
Gain = 6.36
THD = 2.43% @ 2.75W RMS measured with ARTA
2nd harmonic = -34.64dB, 3rd harmonic = -36.62dB ,
4th harmonic = -48.47dB, 5th harmonic = -55.28dB
60 cycle below -80dB, 120 cycle below -60dB
I failed to match the designers benchmark of 1% distortion at 4W RMS.
Perhaps deviations from the original design or using cheap AES PT-31 OPT's is a contributing factor.
That is it for now.
The PS configuration is :
ICT (10 ohms end to CT) 230 - 0 - 230 -> 5V4 full wave -> 35mH/.79 ohm -> 20uF -> 2H/7 ohm -> 100uF -> 1K -> 50uF//200K shunt to ground as a bleeder.
I could bypass the 5V4 with SS diodes, that would raise the B+ some.
The THD at 1W would be a bit lower because there would be a more total power output.
The signal circuit has some minor variations in grid leak values, grid stoppers, driver plate load resistor, but nothing that radically deviant.
My circuit analysis is only the bare minimum of what can be looked at.
I don't understand what are you trying to accomplish with this analysis?
You should keep the LM317 at some distance from the socket to prevent overheating by thermal transfer from the tube ( what is normal for output tubes is way too high for regulators like the LM317).
The cap bypassing g2 to ground is superfluous and definitely not foreseen by the schematics. Remove it and repeat the same set of measurements.
It's useless analysing the circuit with a B+ this low: unless you are looking for a worst case scenario.
Last but not least, your bandwidth measurements are influenced by the output transformer. Anyway, try to stick to the schematics when analysing, than if you want you can modify and post the rest as "worst case scenario with modified (butchered) circuit". The difference between a surgeon and a butcher is in the subject, knowledge, and goal...
******
http://rh-amps.blogspot.com/
I should have called the thread "SE EL84 w/plate to plate local feedback" rather than referencing the RH84 design.
No offense to the designers intended.
The RH84 is a good starting point and is a well documented , popular design with no silly pretensions . He's done all the work already , it's a decent sounding topology with wiggle-room for similar output valves but I wouldn't mess with the basic RH84 driver as it appears to be synergistic . I used EL822 and also tried 6P15P in place of EL84 and got good results .
Al
I get the feeling that Alex does not like people deviating from his published signal circuit values or voltages, LOL.
The power supply was another matter apparently.
A lot of guys are trying this circuit and doing all sorts of things, so my experiments are not that radical.
I had to try the low DCR approach since the parts were at hand.
alex gets a bit cranky but he means well. I work with an engineer like that, sharp and often taken to be a prick..but just not a salesman.
good job on the analysis IMO.
I bet it'd take good iron to get the 1% distortion at 4w, and its probably 3rd harmonic dominant.
power supply influence, maybe...
tubecad article that explains why your ccs did not work without a bypass:
http://www.tubecad.com/2011/08/blog0210.htm
In my view your results are actually satisfactory having in mind that the sim does not take into account a real world transformer. It's more like "this is what you could get, now it's up to you and your choice of OPT".
Plus, 6k is more suitable and was used in the sim, and with pentodes and beam tetrodes primary impedance is more important than with triodes, i.e. there is a "sweet spot effect".
The OPTs you have used are extremely cheap and probably worse than some scavenged or recycled OPTs from old radio receivers. Just compare your results with a standard 2A3 SE amp and you'll get what I mean. Or, compaee it with other EL84 SE amps and you're going to understand why it had so much success.
******
http://rh-amps.blogspot.com/
this rh84 measures better than the last el84 amp I built. I think I ended up around 2.2% distortion at 1 watt using el84 UL with edcor iron... 12at7 led biased... so alex, I Like your design! I think i'll switch my el84 amp to it and give it a listen.
I built a RH84 from a console amp I found next to the road with a FREE sign. I was driving my pickup so I stopped and took it home.
I have a scope, function generator, dvm. Can I measure distortion with only these instruments? I'm more curious than anything.
The amp works great for what I'm using it for. Sorry for the lousy pic.
looking for some jazz and a little libations - joe strummer
Sure a scope can tell you lots, you will be able to visually see distortions. IMO for us amateurs a visual representation is more useful then a static number.Anyhow I won't spend a lot of time on it in this thread, but make sure your scope and DVM can be used at those voltages and don't float the ground.
You should find a suitable power resistor for your OPT. 8Ohm? high wattage.
Have a look at the link, it shows you what phase and symmetry and 10% harmonic distortions look like.
Farther down it shows what noise, parasitic oscillation, clipping and reservoir capacitor ripple look like.It also has a nice section on square wave testing, shows you what the altered squares mean. Also shows the effect of differentiator and integrator (tone controls) circuits.
Take your time, maybe google around a bit, check out youtube this will give you a better feel for what you might be comfortable testing.
Edits: 01/23/15
Thanks. I do have a 50w dale power resistors and heat sinks I use for a loads.
looking for some jazz and a little libations - joe strummer
I have a few other OPT:
Trancendar SE 5K/8 10W, Heyboer 5K/8 14W and Hammond 125ESE.
The only other end users of AES PT-31 have indicated trying them in low power guitar amp circuits.
For biamp with 811B at 1500 cycles up, the AES PT-31 sounded pretty good in the test amp.
Thanks for your input and review of this thread.
I also tried adding the zener diode to drop voltage to g2 from B+.
Instead of the single 21V zener, I used (2) 6.2V Motorola zeners in series and a 22uF cap to ground.
The g2 voltage sits slightly below the plate.
When I get more measurements, they will be posted.
Did you try it again after everything cooled down? The LM317 is supposed to be thermally protected. I have to say, I'm not a fan of connecting a device like the 317 directly to the cathode.
--------------------------
Buy Chinese. Bury freedom.
I added a small bolt & screw to the output tap on the LM317 for heatsinking and the amp seems to be playing normally again.
Normally, I put such circuits on small perfboards with heatsinks and not directly on the tube socket.
I have installed the LM317 on numerous amplifiers including 3 RH types. All were installed directly on the cathode pin of the tube socket with seemingly no problems. The bias was stable in all the amps and so far no problems. I don't have instrumentation to measure its effect but to my ears, the sound is improved especially in terms of dynamics and bass. regards, Dak
I needed to put the LM317 on a tag strip away from the tube socket, then put the cap on the socket to get it to regulate the current properly in my test amp when I raised the B+ up to 300VDC.
Never tried LM317 as a CCS before, so I was on the low end of the learning curve in this case.
The only CCS I have used before were for plate loading of driver stage and finals with BJT or mosfet devices. Never bypassed any of those.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: