| needtubes |
Hello all-
I was looking into a low voltage tube amp, but the more I learn about tubes, the more the low voltage option seemed to diminish. Odds are, I could not get good enough output with my headphones to warrant building. I did a little searching of the archives and came upon ashok's design (I had searched before, but had somehow missed this one), which I think looks fairly nice. Attached is the amp and regulated power supply:
http://www.diyaudio.com/forums/attachment.php?s=&postid=89236
http://www.diyaudio.com/forums/attachment.php?s=&postid=89259
I would use this with a 6DJ8 variant, instead of the ECC83. Does anyone have any opinions as to how this would sound? What about the regulated PS? What voltage rating should my transformer be? Current rating?
Thanks |
|
|
| Sch3mat1c |
Skip the MOSFETs. Just use a nice high-current triode (circa 12B4, there's many with that range of specs). 150V supply is no problem for tubes.
Tim |
|
|
| needtubes |
I am by no means a wiz at biasing or working with tubes apparently, and would prefer to do a tried design instead of work on my own as of now. I worked with the low voltage tubes, and now my power supply tranny is messed up. (don't ask) I also nearly fried a few things...
The need for tubes is killing me... I just want something up and running. If I had the money, I'd buy a kit. But, I cannot afford it. I doubt I could get anything running with another tube and experimental values (I probably wouldn't even get this running). |
|
|
| ashok |
Hi Needtubes,
I am working on a 6DJ8 circuit with less components. It still uses an ss output with a 40 volt dc supply. It has a class A output . Will post the circuit and performance in a day or two.
Am trying to make it so that there is no pcb required.
Cheers.
Ashok. |
|
|
| needtubes |
| 40V supply is a little more in my comfortable range. :) I will definitely take a look at this. |
|
|
| ashok |
Hi Needtubes,
Tubes and low voltage I think is generally frowned upon. But if we can squeeze out decent performance it might be worth the trouble designing one.
My intention was to use a power supply that used capacitors rated at 50 Volts or lower. This keeps costs down. The only low priced low voltage tube that I can get is the 6DJ8 and the 6922.
So when I saw your post I decided to complete my design quickly.
The components have exceeded my target number but are essential for good performance.
Frequency response into a 300 ohm load is -0.5 db at 20Hz and about -0.1db at 20kHz. Distortion is less than 0.1% at nearly 4 volts peak ( 2.8 V rms). That will be about 26.7 milli watts into 300 ohms. With a 96db per mW headphone you will get 110 db!!
The output impedance is also very low - about 1 ohm over most of the bandwidth and rising to about 8 ohms at 20Hz. That I think is quite good.
The power supply is a simple capacitance multiplier. The 5.6 k ohm resistor may have to be chosen to get about 3 volts drop across the collector emitter. This circuit has not been tested yet.
I have collected the parts and should be able to do this in a couple of days. The bootstrap capacitor is essential to improve open loop gain and reduce distortion.
The output is a class A stage with an idle current of about 0.06 Amp. Q2 is a current source with a 9.1 volt zener at the gate.
All resistors are 1/4 watt and only the 3.9K resistor is a 1/2 watt type. Q1 and Q2 will have to be mounted on a simple heat sink. They don't dissipate equal amounts of heat.
Cheers.
Ashok. |
|
|
| needtubes |
I like this circuit. Shouldn't be all that expensive to build, and if it sounds good, should be well worth it. :)
R8... the 300-ohm resistor- I am assuming this is the load, correct?
If I want a volume control, can I just add a 50k pot in front without changing anything?
Thanks |
|
|
| ashok |
Yes R8 is the load. Actually it should have been a large value resistor say 100K . This would permit the output capacitor to charge even if the phones are not connected. Otherwise you would get a bang when you plugged the phones in .
So if R8 is 100k the phones load is not shown ! It will be in parallel with R8.
You can use a 50 K pot for volume. That's what I intend using.
Cheers.
Ashok. |
|
|
| Sch3mat1c |
Is Q3 what's known as a capacitor multiplier? Uh I think...
Tim |
|
|
| ashok |
Hi Tim,
Check out http://sound.westhost.com/project15.htm
This describes the capacitance multiplier.
Q3 is the capacitance multiplier device. The collector base
resistor should ensure that you have enough voltage across the transistor collector emitter to keep it in the active zone - above 1.5 volts under all supply voltage conditions.
Cheers,
Ashok. |
|
|
| fdegrove |
Hi,
| quote: | | This would permit the output capacitor to charge even if the phones are not connected. Otherwise you would get a bang when you plugged the phones in . |
Don't you mean that the function of R8 is to PREVENT the coupling cap from charging when no load is attached?
| quote: | | Is Q3 what's known as a capacitor multiplier? Uh I think... |
Yup. And you can just as well do that with a tube too.
Why it isn't used more often is beyond me, it offers a lot of advantages for very little outlay.
Cheers,;) |
|
|
| ashok |
Hi Frank,
R8 is a large value resistor which permits a dc path to ground. The capacitor will have to charge up to the dc value between output and ground. This ensures that there in no dc on the output when the phones is plugged in. If R8 is missing , the capacitor will charge through the phones and cause a loud transient crack in the headphones - if it is large enough it can blow the headphone coil and diaphragm.
BUT, I am not sure at the moment how fast the dc voltage builds up at the output. If it is slow enough this problem will be reduced.
The capacitance multiplier itself has and RC value of about 1 sec or so. Let me check this . I had overlooked it.
About the capacitance multiplier.
The effective capacitance = hfe x capacitance value.
With a single transistor we can easily have a x 125 factor. With a Darlington we can get a factor in excess of x 1500. Maybe a tube will not be able to match this easily. I have not looked at it -- but maybe!
Cheers,
Ashok. |
|
|
| fdegrove |
Hi,
| quote: | | R8 is a large value resistor which permits a dc path to ground. The capacitor will have to charge up to the dc value between output and ground. |
O.K., we're talking about the same thing...just expressing it differently.:)
As far as the value of R8 is concerned I think 100K should be fine.
What I usually do is add a mute switch that shorts the output to ground while the circuit warms up besides having the bleeder resistor.
| quote: | | With a single transistor we can easily have a x 125 factor. With a Darlington we can get a factor in excess of x 1500. Maybe a tube will not be able to match this easily. I have not looked at it -- but maybe! |
The situation is different when a cap multiplier is used in regular high voltage tube circuits.
No need to tell you that a multiplication factor of *100 at 200 to 300V is massive as far as Joules are concerned.
While you could use a penthode with a mu of 1000+, it won't have much practical use and will quite possibly add too much noise to the psu.
Cheers,;) |
|
|
| ashok |
Hi Frank,
You are quite right about the massive increase in Joules in high voltage circuits. I should try out a tube multiplier one of these days - for an all tube circuit.
Thanks for pointing this out.
Cheers. |
|
|
| ashok |
I checked the dc voltages. The dc rail comes up rather slowly , about two seconds or more. In addition the dc at the output seems to have one spike less than a volt in amplitude and a bump less than 0.2 volts that settles down in a few seconds. Looks like the phones may not be stressed after all during switch on.
BUT the resistor R8 (100K ohms) must be there. If the output is left open and the system switched on the output dc will go up to about 20 odd volts and the output capacitor will not be charged. When the headphone is plugged in the capacitor will charge up to the 20 odd volts through the phones and cause a nasty 20 volt spike in the headphones - something that we want to avoid.
Cheers. |
|
|
| needtubes |
Hello-
I have some questions about parts selection.
I plan on using fairly high quality components- Elna Cerafine caps, Orange Drop caps, Vishay/Dale resistors, etc. However, I am quite limited on the Elnas. What voltage rating should C3 and C4 be? The only Elnas I have access to are 100uF, 100V and 100uF 35V for those. Will the 35 V be ok?
Can I use a 16V cap for C2?
Will .47uF be large enough for the input coupling cap? That is as large of an Orange Drop as I can get. Is a polypropylene Orange Drop better than a Solen metallized polypropylene cap?
Should the 22k resistors be 22.1k or 22k? I can get either.
Thanks |
|
|
| ashok |
Hi Needtubes,
All resistors are 1/4 watt except R6-3.9K which should be at least 1/2 watt. It dissipates about 0.3 watts. A 1 watt will be cooler but I don't think it should matter if you use a 1/2 watt device.
Capacitors: C2 can be a 16 volt device.
C3 and 4 can be 35 volt devices as they have much less voltage across them in operation. C5,6 and 7 are supply caps operating at 40 volts or so and should be rated at least 50 volt.
I am going to use 1N4007 x 4 as the bridge rectifier. The heater also will be run on 6 volt ac . I will solder a 470 ohm resistor to each heater terminal. The free end of the resistors will be taken to circuit ground.
You can add a snubber cap - say 0.01uF 100V across each diode.
The supply current per channel is about 70 mA .
C1 can be 0.47 uF 63 volts. THe -3db frequency atthe input is about 7Hz with a 1uF cap and about 15 Hz with a 0.47 uF cap.
This might be quite OK in practice. Check it out in practice.
With C1=0.47uF and R01 changed to 39K and R9 to 150 K you should have about the same LF response. The 39K will degrade the S/N ratio as it is in series with the input. But I have not checked by how much. Probably not enough to cause any trouble as this is not really a pre amp stage. Best method is to try it out.
About the Orange drop and Solen parts. As far as I know all pp and polyester caps are made of metallised plastic film. I find the Orange drop capacitors very good and I remember seeing a report somewhere on the web that the lowly ( cost wise) Orange Drop caps were tested to be as good audibly as some other exotic caps. I would go for the Orange Drop without a second thought.
I should be able to work on my circuit today. We should compare results. My caps are all from Philips.
About 22k or 22k1 parts. Use anything you want. Make sure that both channels are identical. If any parts are reasonably critical , they should be R5 - 1Kohm, R7-82 ohms, and R3 - 18 Kohm. Remember that R8 is 100 Kohm ( anything from 10K to 100 K ohms is OK ) and the headphone output is parallel with this resistor.
Cheers. |
|
|
| needtubes |
Any news of how the project is going? I have been unable to gather components for lack of time and money. (poor student... the usual complaint)
I am interested in the results that you have gotten. |
|
|
| ashok |
Hi Needtubes,
The amp has not yet been soldered. I am off to collect the transformer today. Here the component shops are just 6 Km away from home. In fact everything is within a radius of about 15 Km from home !
Hopefully it will be ready tomorrow. Will keep you posted.
I am making a pcb because it is really hard to make a point to point connection neatly. I will have a spare board. If it works well I can mail one board to you.
Cheers. |
|
|
| MIKET |
| This circuit looks like a winner, hope to hear back from you as to how it sounds.:scratch: |
|
|
| ashok |
Hi,
I defaulted my stated time of completion. Been very busy with real work !
I finally completed a pcb design for this - a 4x4 inch board. Has everything on it except the input/output sockets and the transformer. If I complete the board (hand painted!) this evening , I will try to post some pics and maybe some test results.
The board has several jumpers. Had no time to refine the design.
Next week is a very busy week so I must try to get it done now.
Cheers.
Ashok. |
|
|
| ashok |
The board is ready . The power supply however has a problem. I am getting only 17 volts in place of 40 volts. The 2200uF cap is heating up. A quick look shows everything is OK but something must be wrong. Will look at it in the morning.
I will try to attach a picture here.
Test results as soon as the power supply is fixed.
Cheers.
Ashok. |
|
|
| ashok |
Hi Needtubes and others,
I got the phones amp to work this morning. The 2200uF/50 V power supply cap was heating up fast enough to blow up. It did not measure as a short or show low resistance. I replaced with another one of the same type. Same problem !! So I went into the market and got a Samsung(?) 4700uF/50V cap and it works fine.
I am attaching the frequency response and the distortion curves.
I played music through the device and used a Sennheiser HD580. It sounds very good. For the outlay I think it sounds great. Great bass and clean all through to hf. The transients are great also.
Go build one. You won't regret it. Some suggestions from me. The heat sink shown in the picture posted earlier is enough for a quiescent current of about 14mA through the MOSFET's. I tried 60mA also and that will need fins. It gets pretty hot - but still touchable if no fins are added ( at 60mA).
14mA is enough for a Sennheiser ( 300 ohms and 97db/mW). But for less sensitive phones and lower impedance you need more current. The limiting factor is clipping on the bottom of the waveform when the current source goes dry. Into a 22k load the amp gives just under 28 volts peak to peak.
The distortion rises rapidly at the point where the load current approaches the current source current. So to get more output you need to up the quiescent current ( for lower headphone impedances).
With 14mA it goes very loud and also sounds very clean. The input capacitor is 0.47uf . You can eliminate this if the signal available is dc free. This may clean up the transients and HF even more.The graph shown has a 1db scale per indicated division.. As you can see it is just about -0.8db at 20Hz.
I could not hear any hum or hiss at full volume .I will try this out when it is absolutely silent and on a HD475 which is quite clinical.
If anyone wants more details I can make it up into a pdf file and email it to you including the pcb layout.
Cheers. |
|
|
| ashok |
Please note that you could get more output voltage by increasing the quiescent current in the mosfet's by decreasing the resistor in the source of the bottom MOSFET. I used 390 ohms for 14mA current. The voltage across it is about 5.5 volts with a Zener of 9.1 volts. With an 82 ohms resistor the current is above 60mA.
You can improve the Lf response by increasing the input cap. I personally prefer to remove the cap. I will try this tomorrow. Small dc offsets from the source will possibly not matter much at all.
Cheers. |
|
|
| ashok |
I decided to implement some changes to see how the amp sounds.
First of all with a HD475 phones I can hear no hum or hiss at full volume. It also sounds wonderful right across the band. The whole unit is pretty inexpensive to build and there are no exotic parts . The input cap is an orange drop 0.47uF/100v cap. The resistors are all carbon film and electrolytic capacitors are Philips / Samsung and a couple of local brands. The tube is a Sovtek 6922 .
The transformers used are 15-0-15 / 1 A and a 6-0-6 /1A type.
I remember noticing that many members consider the bypass capacitor on the cathode resistor to be sonically bad. So I removed the cap after the initial tests. I feel that the resistor sounds better without a bypass capacitor. It looks like low level detail comes out better. An example is the sound of a bell. Without a capacitor I can hear the decaying sound 'very' clearly.
I must put back the cap to compare again and make sure I am not imagining the difference! The mind can play tricks!
Would anyone care to try out this mod and comment on it.
Anyway my final circuit will not contain the bypass cap.
It sounds great and I think it will also be a very good single stage (inverting ) pre amp also. It will be able to handle any type of cable on the output as the load is handled by the MOSFET'S. I will put up the response of the amp with the input cap removed.
If anyone can try out 'exotic' parts , please let us know how it works out.
Cheers. |
|
|
| ashok |
1. I removed the input capacitor . So now the response is ruler flat up to about 10KHz and drops about 0.2 db at 20 KHz.
2. I increased the quiescent current to 44mA as my other phones are less than 100 ohms.
The unit sounds great. You can hear very small changes in a mix - something I never heard before. Voice is great and I am really impressed with the amount of low level detail I can hear. I am really glad I built this unit.
Thanks to Needtubes for pushing me to get it done and post the results. It speeded up the process of building it tremendously. So Needtubes can have the spare board that I have. Email me you address and I will have the board mailed to you - if you want the board.
The distortion curve with 44mA is attached. As you can see the overload point is much higher than before. I forgot to mention that the output impedance is below 0.01 ohm over most of the band and rises to about 0.04 ohms at 20 Khz.. That's real low ! This unit could also be used as a pre amp I guess .The noise floor seems to be below -90db ( ref 1 volt output) as seen on the FFT. Voltage gain is just under 4 . That should be enough for most applications.
Cheers. |
|
|
| ashok |
I have attached the FFT made at a quiescent current of 44mA ( through the MOSFET's) and an output of 1.45V.
Load was 300 ohms.
The noise level as seen on the FFT is below -90db ( with reference to this signal the S/N should be about -93db ).
I plan to plug in an ECC tube and see what results I get !
Cheers. |
|
|
| ashok |
I just plugged in an ECC81 ( NOS Mullard ) and a ECC82 (RFT).
The ECC82 had slightly higher distortion than the 6922.
The frequency response is -0.5 db down at 20 Khz. The 6922 was just about -0.2db at 20Khz.
The ECC 81 was surely poorer. The distortion was significantly higher and the frequency response was -1db at 20 Khz.
Frankly I never even expected so much out of these tubes at such low voltages. The supply voltage is just about 37 volts with anode voltage at about 20 volts.
I must try out the units with headphones and see how they differ while playing music. The good news I guess is that one can get away with an ECC82 also. This is a LOW cost headphone amp and performs very well.
Cheers. |
|
|
| fdegrove |
Hi,
| quote: | | The supply voltage is just about 37 volts with anode voltage at about 20 volts. |
At low plate voltages such as the ones you're using it comes as no surprise that the 6922 comes out best.
It's still relatively linear whereas the others aren't.
Cheers,;) |
|
|
| needtubes |
Hello-
Life became busy for me recently.... so I have not posted as much.
I am glad your results turned out well. :) Perhaps I should begin collecting my parts... I think I have some laying around already. My cans will be the HD580s, so everything should work well.
EDIT: I cannot seem to email you- it says I don't have permission. :confused: |
|
|
| ashok |
Hi Needtubes.
Your email is disabled too !
You can email me at
spam_ashokm@sify.com_spam
Remove the underscore (_) and the (spam).
I get so much junk mail that I am afraid to put up the email id by itself.
Cheers.
Ashok. |
|
|
| ashok |
Hi Frank,
Yes the 6922 is suited for low voltage applications. I however did not expect any of the ECC series to be anywhere near the 6922 as the voltages were so low.
Though they do have much larger distortion figures than the 6922 it might appear that they "might" sound acceptable at low volume.
Well my listening test confirmed that the 6922 is far better than either ECC tube that I tried. The ECC82 is an acceptable substitue if one does not have a 6922/6DJ8 at hand. But the 6922/6DJ8 will have to be used eventually if one plans to use this as a headphone amp regularly.
I found the sound of the 6922 easier on the ear and cleaner than the ECC82. The ECC81 does not even figure over here.
Iused a Marantz CD67SE-II that I modified , as the source.
Transients are really great on a HD580 and HD475. The HD 580 has of course more low end than the 475. The 475 however sounds "faster" especially with transients.
The circuit has only one capacitor in the signal path. The output capacitor - a 100 or 220 uf type. I don't have a Black Gate to see if there can be any difference between that and the Philips type that I have used. If used as a preamp however , I can change this to a 1uF polyester type - improving the audio path ? In fact , maybe I can connect both a 100uF and 1uFPolyester and take separate outputs to a headphone socket and the other to an RCA socket ( for use as a pre amp). Is this a good idea?
Cheers. |
|
|
| fdegrove |
Hi,
| quote: | | If used as a preamp however , I can change this to a 1uF polyester type - improving the audio path ? In fact , maybe I can connect both a 100uF and 1uFPolyester and take separate outputs to a headphone socket and the other to an RCA socket ( for use as a pre amp). Is this a good idea? |
Certainly, you could do this.
But why polyester caps?
Generally speaking any metallised polypropylene cap should be a better coupling cap.
If you can't find those the metallised polycarbonate one are second best.
Just don't load your headphone/preamp with both the headphones and amp at the same time.
Cheers,;) |
|
|
| ashok |
Hi Frank,
Thanks for you post. I was generally mentioning about caps and didn't really mean polyester specifically. I just typed what came to mind. Film versus electrolytic.
Here we do have a problem getting what we want , though generally plenty of 'general' components are available.
We do get polycarbonate capacitors and I have used a few. I am not sure about their sonic capabilities against the 'better known' brands. We don't have much of a choice.
The 1uF that I used is from Philips and is marked 1uF/K/250V/MFC. The other types have axial leads and need more board space and worth it only if they are better. I haven't really compared them. Maybe I should.
Have you any experience with (orange) Philips capacitors like the 1uF I mentioned? ( like the Sprague Orange Drop ?)
Cheers.
Ashok. |
|
|
| fdegrove |
Hi,
| quote: | | Have you any experience with (orange) Philips capacitors like the 1uF I mentioned? |
I know which ones you mean but I've never used them for audio projects.
Maybe someone else has?
Cheers,;) |
|
|
| ashok |
I did a hasty test as a pre amp.
First I played the CD player direct to an Alps Blue pot and then into the power amp. Then I played the same music through the new 6922 phones amp. Cd player to Volume pot to 6922 amp. Output of 6922 amp to power amp ( input pot at max position).
It sounds good with a few reservations. Bass is a bit smeared. It sounds 'a bit' wooly in comparison. Mids and hf seem to be very good. It is 11pm here so I can't turn up the volume. Will do that tomorrow. I will also remove the otput cap on the phones amp
( 220uF at the moment ) and use a better quality film capacitor. Maybe the bass will improve. The bootstrap capacitor 'could have been' a Black Gate ! I will also try high speed diodes in the power supply and also a tube rectifier ( later).
For the outlay I think this has worked out very well. I am especially pleased that there is no audible hum or hiss - a big relief. Now I can work on the circuit and refine it some more. It should be hard to beat at the price.
BUT compared to my new tubed passive eq RIAA amp and vinyl playback system , the CD player sounds FLAT ! There is so much "more" information on vinyl. After hearing it for a few weeks now , so much seems missing. I have CD versions of vinyl discs for comparison. What a diff. !
Cheers. |
|
|
| ashok |
Hi Frank ,
The MFC caps are metallised polycarbonate I think. Philips has another set of codes on their caps shown on the web. KC stands for polycarbonate and M for metallised. So I guess this must be polycarbonate.
I tested the amp at fairly high volume and the difference between the 6922 and ECC82 becomes very apparent. But then it is not fair to compare the ECC82 when it is being run at a voltage where it was not meant to function. I must try an ECC82 with a 250 or 300 volt supply and see how that will compare with the 6922. Would you have any idea?
6922 at about 20 volts anode voltage ( +40 Volt supply ) and
ECC82 at about 125 volts anode voltage (at +250volts supply ).
Cheers.
Ashok. |
|
|
| fdegrove |
Hi,
| quote: | | The MFC caps are metallised polycarbonate I think. |
That's what I think as well but often the F stands for foil.
In this case I'd doubt it but you never know.
Metal and foil caps are even better than plain metallised ones so, maybe you're in luck.
For the record:
M = Metallised.
K = ?
T = Polyester.
C = Polycarbonate.
P = Polypropylene.
That's just the most common ones.
| quote: | | I must try an ECC82 with a 250 or 300 volt supply and see how that will compare with the 6922. Would you have any idea? |
To be honest I wouldn't even bother with the ECC82 their distortion spectrum is nothing to write home about.
Rather I'd rework the 6922 circuit so it runs from approx. a 200 VDC supply and revise the bias settings.
This should give you more dynamic range and hopefully even less distortion.
Cheers,;) |
|
|
| ashok |
I had a couple of people asking about the headphone amp circuit.
I think it is linked in this thread.
http://www.diyaudio.com/forums/show...d+headphone+amp
(post#6)
About running at high voltage and low voltage.
I'll try to check that out and post the results. The board is somewhere around. I'll have to look for the pcb layout . It's not on this computer for sure.
One catch about higher voltage operation is the power dissipation of the output stage . We need the current for low impedance headphones but do not need the extra voltage possible. Maybe some mods are possible. Let me work on that ......later.
Cheers,
Ashok. |
|
|
|
