| croccodillo |
Hi all,
I'm trying to build a PSE amp using two paralleled KT88 for each channel, driven by a 12AX7 with the help of a PowerDrive circuit (Thanks to TubeLab).
I have a lot of surplus transformers, but no one with a suitable voltage for the KT88s heaters (6.3V).
I have instead some 12V transformers, so my thought was: can I place the two KT88 heaters of each channel in series (in a string)? This should give me a 12V string, and I could use some resistor, in parallel with each heater, to equalize the two voltages across the two heaters themselves.
Is this feasible?
Thanks,
Giovanni |
|
|
| Giaime |
Obviously you can! Better would be to rectify the heater supply and regulate it to 12.6V precisely. Or better yet, with a CCS as I've seen. But this may be overkill... I think that simply putting them in series without any resistors and tricks will do good.
Btw, Giovanni, are you a member of some italian forums? I don't remember to have seen your nick anywhere else.
Edit: could you provide a schematic of your amp? Are you driving the KT88 in A2 with the powerdrive? |
|
|
| Shoog |
Series should be fine- no resistors necissary.. Running them slightly lean wont be a problem. Don't bother going DC, there is absolutely no advantage with KT88's.
Just make certain that your transformer is beefy enough and wont sag under the strain.
Shoog |
|
|
| dhaen |
| I would add the caveat that both valves be of the same manufacturer and year. |
|
|
| croccodillo |
| quote: | | I would add the caveat that both valves be of the same manufacturer and year. |
Yes, I have a matched quartet from JJ Tesla, coming from the same lot of production (at least this is what they told me...).
Giaime:
I have no schematic of my amp yet, I'm still trying different configurations and building some parts before to put all togheter.
However my amp will be a PSE (two KT88 for each channel), with para-feed output topology feed by a SS CCS, a PowerDrive to drive the two KT88s (modified in manner it can equalize continuously the current into the two paralleled tubes) and a not-yet-decided input stage (my favorite input stage at the moment is a long tail with a CCS).
The KT88s cathodes will be connected to ground, while the PowerDrive will provide the needed negative bias voltage.
Some sort of local feedback will be added, from the output stage to the input one (no feedback from the OPT, due to phase shift).
The B+ will be about 600V, regulated by an high voltage mosfet.
The PowerDrive will be powered by two voltage (Bias+ and Bias-), of about 24V and -120V (both of them regulated with mosfets).
This should permit me to drive the KT88's grid from about +20V to -80V, going into A2 zone.
I have built the four CCS (two with a current of about 5mA -input stage-, the others two with a current of about 150mA -output stage-).
At now I'm building the power supply, next will be the two modified PowerDrive boards and, at the end, the amp itself.
Ciao,
Giovanni |
|
|
| richwalters |
What cheeses me off is stupid images been put on thread pages with an imbedded virus been put on before virus system updates.
Moderators wake up and stop this stupid time wasting practice.
richj |
|
|
| Giaime |
| quote: | Originally posted by richwalters
What cheeses me off is stupid images been put on thread pages with an imbedded virus been put on before virus system updates.
Moderators wake up and stop this stupid time wasting practice.
richj |
What are you referring to Rich? :confused:
Giovanni: I never heard of running KT88 to A2. Can you point me at some links of someone who did this? How much power are you expecting to get from this PSE? Maybe 50 or more watts... |
|
|
| dhaen |
| quote: | Originally posted by richwalters
What cheeses me off is stupid images been put on thread pages with an imbedded virus been put on before virus system updates.
Moderators wake up and stop this stupid time wasting practice.
richj |
:cop: Rich if you know of any on diyAudio.com please let us know their whereabouts and how you detected them. |
|
|
| richwalters |
Although I'm implying Gaime got this on his system...(I don't want a fight with a nice fella) but the attachment http://www.danasoft.com/sig/Croccodillo.jpg contained a bug that upset my computer clock. If I'm wrong Ill take it all back but the forum mail was the only one I had in the "in box". I had opened other diyaudio threads, perhaps the villan is elsewhere.
richj |
|
|
| Giaime |
| I still can't understand. :confused: |
|
|
| dhaen |
Rich,
I can see you are not blaming Giaime or any particular person, but are suffering some indignation from having your system compromised.
Presumably you are thinking about the "wmf flaw", which was not patched at the time of the posts. I don't see what anybody (apart from Microsoft) could do about this, if indeed there was malicious code in the file. Searching for "Danasoft" and "wmf" or "exploit" returns no pertinent results in Google, so I doubt there was.
External and embedded web links do have a potential for misuse, but the buck must stop with the receiving system. No system is invulnerable, but some are a lot better than others.;) |
|
|
| Giaime |
If you're referring to the image in the signature, that's Giovanni's one, not mine!!!
I now understand, that penguin image has got a virus in it? I will check with Norton if my system has been infected. |
|
|
| richwalters |
Oh well apologies for the glitch... unwanted distraction from 88'topic.......some things are virtually beyond our means. However we all have to be on guard.
Back to the series heater. I spotted an abnomally with the 6550 series......not all vendors have identical hot resistances. I've some drawing 1.3A and some 1.8A at 6.3V.
Do the gizmo with ohms law and you will find at 12.6V one stews and the other runs a chill beyond the lower 10% tol..
richj |
|
|
| croccodillo |
Richwalters:
I apologize if my "signature" has caused problems to you... But as far as I know Danasoft and its products (my signature is one of them) is a secure software house.
And my firewall/antivirus/antispam server does not detect anything from it.
| quote: | | Giovanni: I never heard of running KT88 to A2. Can you point me at some links of someone who did this? How much power are you expecting to get from this PSE? Maybe 50 or more watts... |
I never heard too.
I'm a newbie, I'm trying to build my first amp.
The output power is not so high, I think I will have not more than 10-15W... I do not need a lot of power, and I do prefer to run the tubes well below their limits in manner to have lower distorsion.
Ciao,
Giovanni |
|
|
| Giaime |
If you want long tube life then forget about class A2, please! ;)
You'll get about 10W in triode mode, that's the best way to use such big tetrodes. Or go ultralinear, or go in tetrode mode BUT only with NFB and regulated screen supply.
IMHO the simplest (and best sounding) configuration is plain triode, g2 connected to the plate. |
|
|
| croccodillo |
Well,
THIS is a preliminary schematic of my amplifier.
It is really simple.
As you can see there's a long tail input stage followed by a para-feed output stage.
The output tubes are connected as triodes.
There are also two SS CCS, and a PowerDrive interstage.
Each CCS needs a separate supply (9V, 10mA), but a 9V 3VA transformer costs pratically nothing, about 2 Euro.
THIS is the PowerDrive schematic, and THIS is the CCS schematic.
THIS is a gallery of my first prototype.
You can see one 150mA CCS installed onto the large heatsink.
The output tubes was driven directly by my laptop soundcard output (no preamp stage), so the sound was not really loud.
The first prototype did not use PowerDrive, next step will be to build two PowerDrive PCB (i have already drawn them) and try them in the prototype.
Ciao,
Giovanni |
|
|
| tubelab.com |
Operating a tube in A2 does not cause any shortening of tube life. Pushing a tube beyond its published ratings does. It is possible to operate a tube beyond its safe ratings in A1 or A2. An amplifier (using normal receiving tubes) can be configured to allow A2 operation. It will spend 99% of its time in A1. The only difference is that the driver circuit is designed such that it can drive the grid of the output tube positive on transients without distortion. The reason for this is to remove one of the common (but not well understood) sources of distortion in a tube amp. PowerDrive was conceived to eliminate this distortion, the extra power output and improved sound were side benefits.
Go to http://www.aikenamps.com/
Select Tech Info, then choose Advanced, then scroll down to What is "Blocking Distortion" for a full explanation. This site is about guitar amps, but the tech info is excellent, well written, and applicibable to all tube amps.
Most receiving tubes were never designed with positive grid operation in mind. Some are linear in this region, and some are not. I have not tested a KT-88 for operation with positive grid bias. If a tube is linear with positive bias, then you can adjust the amplifier to take advantage of this and gain some extra output power. If a tube is not linear in the positive grid bias region, a driver designed for A2 operation will still remove the distortion that occurs when the driver clips, and the severe distortion that occurs when a transient peak upsets the charge on the coupling capacitor which then takes time to recover (blocking distortion).
I have used A2 successfully on many amplifiers (but I have not tested a KT-88). The tubes that benefit the most and can provide extra power ( all pentodes tested in triode mode):
300B, 845, 211, 2A3, EL-34, 6LW6, 6AV5, 807 and the 833A
The 45 didn't provide any extra power with A2 capabilities, although the amp did sound much better with PowerDrive.
Some tubes were designed to operate with positive bias on the tube these REQUIRE a driver capable of A2 operation. The obvious ones are the 811A, the 3-500Z, and the 838. |
|
|
| richwalters |
| quote: | Originally posted by croccodillo
Well,
a PowerDrive interstage. |
Ah! deserter...... he's using solid state in the driver stage...I've always suspected this stage in any amp determines alot of the quality aspect more than the output tubes. I wouldn't be suprised after auditioning he does an about turn away from mosfets. Anyone else using such SState driver applications in thier designs ?
omo richj |
|
|
| Giaime |
| quote: | Originally posted by tubelab.com
Go to http://www.aikenamps.com/
Select Tech Info, then choose Advanced, then scroll down to What is "Blocking Distortion" for a full explanation. This site is about guitar amps, but the tech info is excellent, well written, and applicibable to all tube amps.
|
I knew it, thanks George ;)
Giovanni, do you really need that big heatsink? :bigeyes: |
|
|
| croccodillo |
| quote: | | Giovanni, do you really need that big heatsink? |
Probably that big heatsink is not big enough....
The heatsink is where the two output stage CCS are mounted.
Give a look at the schematic: the amp is a para-feed, the B+ voltage is about 550V (I'm not sure if raise it up to 600V).
The output tubes will sit at about 350V at idle: that means that the CCS feeding the KT88s will work, at idle, with a voltage of about 200V and a current of 150mA (for each channel).
Thus, each CCS will dissipate a power of about 250V*0.150A=30W.
Having a stereo amp the two CCS will dissipate about 60W of power.
As you can see, I really need a big heatsink!
I made some calculation, trying to simulate the heatsink (I have no data about it, it came from a surplus stock), and it should handle the power and raise up to 60°C at idle...
It could seems a big amount of power, but the four KT88s will dissipate about 100W of power (not including the heaters)!
Ciao,
Giovanni |
|
|
| Giaime |
| Yes, I see, you need it big. But why so high B+ voltage? With 400V you could have done just well... |
|
|
| croccodillo |
Somebody told me that KT88 works better with high voltages...
And using 550V I can have an output voltage swing of almost 400V (from 150V to 550V)@150mA.
Moreover, the tubes, when in idle, sit in the middle (I do not know how to explain better the concept), at 350V, and can go down to 150V (-200V from idle) and up to 550V (+200 from idle), reproducing a (ideally) perfect signal at output.
If I'm right this should lower the distorsion and increase the output signal correctness (sorry for my bad english explanation...).
Again, this should also increase the damping and improve the sound.
Ciao,
Giovanni |
|
|
| Giaime |
Oh yes I see. But the rise in plate voltage in a tube amp is derived from the inductive loading of the output transformer, not necessary from the power supply. I mean, with a 400V supply, plate voltage can go up to 600/700V because of the inductive loading, that's not actually necessary to use 700V supply. Even with parafeed topology you're using.
Obviously it won't hurt to waste a few volts ;)
Correct me if I'm wrong... |
|
|
| croccodillo |
I'm a newbie with tubes, and I'm still learning how to use them.
I was born as a "sand" guy, and I'm used to think in that way: that means that my solutions derive from sand topologies, and my mental simulation of circuits has a sand base...
Thus, your observation make me think, since up to nw I never considered OPT inductive load voltages: thanks a lot!!! |
|
|
| Giaime |
I'm a newbie too Giovanni, but I made a little simulation on TubeCAD (very very good program, go purchase it immediately): a PSE with KT88 (not parafeed): the results with a 350V power supply are that plate voltage can go anywhere from 1 to 678V, with an average of 265V. So you have to think with tubes in mind ;)
I'm not a sand-state expert, but I think that a SS output stage can put on the load a maximum voltage equal to the power supply voltage minus a few volts, with tubes that's very different.
I think you have to choose a proper operating point for your KT88, then choose a B+ voltage of some volts higher the quiescent plate voltage (IMHO 400-450V not more) and let the OT do the rest :D |
|
|
| dhaen |
| quote: | | I think you have to choose a proper operating point for your KT88, then choose a B+ voltage of some volts higher the quiescent plate voltage (IMHO 400-450V not more) and let the OT do the res |
Sounds like a good starting point to me:) |
|
|
| SY |
| quote: | Originally posted by richwalters
Ah! deserter...... he's using solid state in the driver stage...I've always suspected this stage in any amp determines alot of the quality aspect more than the output tubes. I wouldn't be suprised after auditioning he does an about turn away from mosfets. Anyone else using such SState driver applications in thier designs ?
omo richj |
I do. MOSFET source followers using IRF820 to drive the 6LF6 outputs. And a proto using a DN2540 cascode source follower. Works a charm. |
|
|
| Giaime |
| quote: | Originally posted by SY
I do. MOSFET source followers using IRF820 to drive the 6LF6 outputs. And a proto using a DN2540 cascode source follower. Works a charm. |
Me too Stuart, and I plan on a PowerDriven EL84 PP when I'll finish the preamp ;)
Let's avoid some of the rigid "tubes or nothing" rules... IMHO sand state stuff, slaved to the Glory of Vacuum Tubes, made to Serve them, is a very good idea. It keeps the intellectual superiority of vacuum tubes with many of the advantages of sand :D |
|
|
| tubelab.com |
First of all if you use a CCS to supply the voltage to the plate of the output tube, the OPT is not in the DC path. The inductive kick of the transformer can not supply the positive swing above the supply voltage. This is known as "chokeless parafeed". Since there is no choke or other inductance in the plate DC path, the head room must come from the power supply, and yes you need a lot of voltage. I had a similar design running from a 400 volt supply and I was only getting 3 watts. This was discussed heavilly in the following thread by most of the same contributors. There is a picture of my test circuit there.
http://www.diyaudio.com/forums/show...&threadid=67437
| quote: | | Ah! deserter...... he's using solid state in the driver stage |
| quote: | | I wouldn't be suprised after auditioning he does an about turn away from mosfets. Anyone else using such SState driver applications in thier designs ? |
This was overwhelmingly the initial response that I got when I put the Powerdrive concept on my web site. I even got hate mail saying that I should change my name to Transistorlab! Now 2 years later, after people actually tried it, that is all gone. The people who have actually listened to PowerDrive like it a lot. It has become the bigest source of questions in my email. Fortunately I am in the process of expanding the how to material on the PowerDrive page to make it easier to understand, and adapt to your own design.
http://www.tubelab.com/powerdrive.htm
A major update will be posted in a few weeks. |
|
|
| Giaime |
| quote: | Originally posted by tubelab.com
First of all if you use a CCS to supply the voltage to the plate of the output tube, the OPT is not in the DC path. The inductive kick of the transformer can not supply the positive swing above the supply voltage. This is known as "chokeless parafeed". Since there is no choke or other inductance in the plate DC path, the head room must come from the power supply, and yes you need a lot of voltage. |
Oh, excuse me George, I was wrong. I didn't think that the OT couldn't be on the DC path, because of the coupling cap. So if you want 600V max at the plate, you need a 700 or more volts supply if you use a CCS, right? That's quite exhagerate for me, much better a CCS at the cathodes instead, and conventional output tranny. You can even save a C in the signal path (and that's very good)...
I guess that if Giovanni has chosen this, there are good reasons. I didn't follow the original thread, so I don't know.
:smash: |
|
|
| Shoog |
I'am still a little concerned about those CCS's sitting on a large heatsink which will be exposed to the outside world and riding at an unhealthy 500V. If ever anything goes wrong (ie a heatsink to tab short) it could be lethal. I suppose if the heatsink is earthed and the case is safety earthed to a RCCB things should be OK as the amp will trip out the mains prity quick.Alternatively could you get your Mosfets in isolated packages.
Even though you save yourself a few volts of overhead - I still prefer a meaty tube CCS. Tubes are built to shed heat in a way that transistors are not.
Shoog |
|
|
| Giaime |
| What you suggest, Shoog, remembers me a CCS KT88 + 807 you've made some time ago... is still working? How does it sounds? Maybe if you post the design here you'll convince Giovanni of that idea... ;) |
|
|
| Shoog |
I have been working on tweeking it and its sound better than ever. I upgraded the output transformers to Power toroidals (made a huge difference), I also increased the amount of plate to plate feedback (160K down from 470K), added an extra gain stage to make up for the unity gain of my preamp. I tried the plate to plate at 100K but it introduced resonances into the output stage.
Using the same 807's after about 8months of 16 hour days, and the bias point seems rock solid. KT88'a aren't even sweating.
A real stunner - I cannot image me building anything that will sound better so I am tentatively going to say my quest is over.
Shoog |
|
|
| SY |
| quote: | | I'am still a little concerned about those CCS's sitting on a large heatsink which will be exposed to the outside world and riding at an unhealthy 500V. |
Shoog, not to mention the stray capacitance inherent to grounded heatsinks... |
|
|
| Giaime |
What's that, SY? Could you explain?
Are you saying that the mosfet + heatsink act like a capacitor and suck out the sound from the plate of the tube? |
|
|
| tubelab.com |
| quote: | | Oh, excuse me George, I was wrong. |
Not needed. I also ran simulations that showed that the OPT could supply the needed headroom even though it was coupled through a capacitor. I had to actually build the amp to find out that this ins't so. Some voltage is there but there is extreme distortion.
The point of this forum is for us all to learn from each others experiences, so we all don't have to repeat the ones that don't work.
| quote: | | I upgraded the output transformers to Power toroidals (made a huge difference), |
Are we talking about mains transformers here? If so how many VA are the transformers that you are using, and how much power is your amp making. 240 volt primary? |
|
|
| Shoog |
240V primary with 6V secondary. I would estimate that the VA rating is in the region of 120VA. An 807 SE, so we're probably talking 4-5watts output - more than loud enough with my speakers.
I used some 12V transformers and unwound half of the secondaries. I then taped the whole assembly up with insulation tape. Very clean detailed bass and, very smooth sound.
I just wasn't prepared to pay the outragous prices been asked for parafeed output transformers. I don't know if real parafeed transformers sound significantly better, but I haven't got the money to find out. I am more than satisfied with the sounds of my cheap mains transformers. I even went and built another parafeed headphone amp with toroidals as outputs - sounds great !
Shoog |
|
|
| tubelab.com |
I don't like the prices for many transformers either. It seems that most vendors raise the prices for any transformer they make that has the word "tube" in the description.
I have used some Talema toriodal mains transformers with good success, but I was using a 56VA transformer on a 2 watt amp. It did sound good though. I will try some larger mains toroids when I build a proper power supply for the KT-88 - 6LW6 amp. |
|
|
| Shoog |
Sourcing mains transformers in the correct ratios at high VA's means custom wound - which brings cost up again. Thats why I ended up using some that I had and unwinding to spec.
I think transformer manufactures are laughing at what the Parafeed crowd will put up with (anyone for cobolt cores).
Shoog |
|
|
| tubelab.com |
| As long as the audiophile crowd keeps buying............. |
|
|
| pedroskova |
I've been using little Taleema power toroids for the outputs of my parafeed preamp and they work fine...especially for 13 bucks a piece.
Someone on the web measured their bandwidth at ~ 5Hz-150kHz. |
|
|
| SY |
| quote: | Originally posted by Giaime
What's that, SY? Could you explain?
Are you saying that the mosfet + heatsink act like a capacitor and suck out the sound from the plate of the tube? |
Not far off. In most CCS, the drain is the business end, and as a CCS, we want the source impedance there to be as high as possible. The drain is also usually the terminal which is brought out to the case. When you put the device on a grounded heat sink with a nice high dielectric constant thermal insulator (e.g., mica), you form a capacitance to ground from the drain. And it can be a pretty significant capacitance. That severely compromises the desired high source impedance, getting worse at higher frequencies. |
|
|
| Giaime |
Thank you Stuart! Could you quantify the capacitance involved, just to get an idea?
Going this way I'm starting to think that we need a tube doing this job ;) |
|
|
| poobah |
Giaime,
The capacitance is in the tens or hundreds (not very likely 100's) of picofarads. But at these high voltages that represents significant energy storage. |
|
|
| SY |
| I don't know offhand, but I'll go measure one and see. |
|
|
| croccodillo |
Hi all,
I choose this topology for few reasons:
I've been always a fan of Class A amplifiers.
My favorite amplifier, up to now, is the Nelson Pass's Aleph 30 clone I built some years ago.
I never finished it: during a listening session, with the amplifier open on my workbench, I made a bad short-circuit on the left channel board, and the channel literally exploded... I never replaced the channel, too much work.
By the way, the large Heatsink you can see used in my tube amplifer prototype was part of the Aleph 30 clone.
When I decided to build my tube amplifier I searched the web for information, until I found the "parallel feed" topology.
I was already thinking about a SE amplifier with a CCS on top (this is the topology normally used by my previous SS amplifiers), and I fallen in love with "para-feed"; I do not know why, but my "instinct" told me that such a topology was MY topology.
Next has been to decide which CCS to use: a tube CCS needs too much voltage to me, so I decided to go with a SS CCS, that can have a voltage as low as 10V.
In this case using a relatively "low" B+ voltage of 600V I can have, on the KT88s plate, up to 590V.
Ciao,
Giovanni |
|
|
| croccodillo |
About my CCS:
The Drain pin of the mosfet (the one connected to the mosfet case) is connected directly to the B+ rail; that means that if it introduces some capacitance, this will be added to the PS filter capacitance, since it is connected between B+ and ground.
The "business" end of my CCS (the one that goes to the output tubes) is the source one, and the capacitance on this side is near zero.
The heatsink is of course insulated from the mosfet drain, and firmly connected to ground.
In case of a short-circuit between the mosfet case and the heatsink, you simply have a short-cicuit between B+ and ground... so the protection fuse on the PS (I always place protection on PS, I hope you make the same) will blow avoiding any danger.
An example: Imagine a "traditional" amplifier, wired point-to-point: at a certain point the B+ wire disconnect from the output tubes and touch the chassis... What happens? exactly the same than could happen to my amplifier if the mosfet short-circuit to the heatsink.
About output transformes:
I too use toroidal power transformer, I'm used to use such a transformer for my job, and they have an extremely wide bandwidth and frequency range.
For what I can heard, they sound great.
THIS is a link to an italian page with an amplifer using a toroidal transformer as output.
Ciao,
Giovanni |
|
|
| Shoog |
I was thinking that maybe some of the deflection tubes might make a good cheap CCS with a much reduced overhead (say down to 100V). Something like the EL36. Anyone think this will work ??
Shoog |
|
|
| tubelab.com |
| I tested several types of tubes for CCS use. I will put the results on my web site eventually, but the short version is that sweep tubes work well for this. The best tube I tested was the 6LW6. That is the tube I wound up using in my test amp. |
|
|
|
