• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Ralph Power amplifier

Hello

I am wondering if anyone have buildt this amp:

http://www.meta-gizmo.com/Tri/interstage.html


I wonder if the rated 275V at the plates are correct. The amp states to give ~8W, and I think maybe the platevoltage should be 375V.
But then the 200uF PSU cap has a wrong Rating too..
The schematics give no currents, but I think it runs at 60mA per tube, thus giving a 60V bias through the 500R cathode resistor.
Anyone knows?

I really want to build this thingy, so I would appriciate some help :) Maybe it gives 8W in class AB? :(

-Stig
 
Stig,

This is very nice topology. I like it. Some fine tuning may be necessary. This is a personal comment and suits my ploarised views.

The HT winding shown in the power supply appears to be rather high. I hope that the rectified and filtered voltages are correct.

6B4G is a 6.3V heater version of 2A3. Maximum Plate Dissipation is 15W. In Class AB operation I would recommend a maximum Plate voltage not exceeding 300V. At 275V as shown in the schematic, the plate current should not exceed 54mA. I would recommend between 45-50mA per tube, to preserve longer life.

With Directly Heated Triodes, a push-pull topology will reduce hum if you transpose the heater connections to the tubes.

The input tube is well chosen. However 1000uF (bypass capacitor) will somewhat slow down your transient response. The only way to overcome this is to change the operating conditions of the input tube. I would recommend a plate voltage 125V, cathode resistor 100 ohm and bypass capacitor 330uF (Elna Silmic II). Current through the tube will then be around 14-15mA and the bias voltage around 1.4V. This combination of cathode resistor and bypass capacitor will not go down to single digit flat frequency response but will be much more musical.

The other compromise in the circuit is the 7.5k resistor in the power supply. This is intended to reduce the B+ for the 417A. However, this will slow down transients to some degree. I would recommend using the centre tap in the power transformer to derive a second power supply of about half the present voltage. Use choke input filter and perhaps two stages of filtering. This should, along with a reduced resistor value, return (assuming voltage drops across both chokes) the desired plate voltage for the 417A.

By the way, tube rectifiers are not my preferred choice. I prefer solid state diodes instead. If you are not already committed, you should redesign the entire HT supply using two stage choke input filter.

By the way, do you need 8W?

Mohan
 
Thanks for Answering!

I've tried out PSUDII and I came up with ~320V using 60mA per tube. Think I'll go for that.
What really got me is that with the rated 275V, -60 volts, will leave 215V at the plate, so I really wonder if it runns in class A or not...

8 Watts is just about right what I need, so this construction appealed to me.
I am a little bit into tube stuff, so I have redesigned the PSU and input stage, inspired by the Amity amps;
A ECC99 at the input, giving an input sens. of 2,45V. No problem for my Preamp! Cathode BP cap is now reduced to 220 uF (300R Cathod resistor.) The LL1635 is replaced by a LL1660.
Further, I use the 6D22S's to do the HV job, and for the ECC99 PV I will experiment with a couple of 0C3's. (tapped from the input choke).
I will also play around with Ultrapath cathode BP at the outputstage.

I plan to use the Sovtek 6B4G, they can handle ~100mA and 400V !

and please, let the reply's streaming in!!! :)

-Stig
 
Stig,





I wonder if the rated 275V at the plates are correct. The amp states to give ~8W, and I think maybe the platevoltage should be 375V. But then the 200uF PSU cap has a wrong Rating too.. The schematics give no currents, but I think it runs at 60mA per tube, thus giving a 60V bias through the 500R cathode resistor. Anyone knows? I really want to build this thingy, so I would appriciate some help Maybe it gives 8W in class AB?






From the schematic one cannot tell much. The 2A3, sorry 6B4G, is specified to achieve 4 watts for 250V at the plate and bias of -45V which would be class A. So a pair of them definitely should give 8Watts, but it doesn't matter anyway whether you 8 or 6.78, isn't it?





However, as this is a standard PP topology, not a differential one, it will clip soft in AB1 operation and sound almost as good and it can be driven harder into distortion w/o sounding awful. But this is the case even if you adjust it to class A operation and then drive one half a bit into grid current.





The schematic is only hint-giving, no specifics anywhere. When you build it, you are practically on your own, with forum member help, i would presume :) . See that you get rid of cathode bypass caps, sonically they destroy more than adding another stage. They are coupling caps. Why should you pay $$$ for an IS trasnnie inorder to get rid of coupling caps between stages and then have them in the ground path?





There is no reason for not using a PP driver stage and an input transformer (except the almost unhearable protests of your wallet) but sound will become even more glorious with 3 trannies in the signal and the driver can be made more juicy. The 417A/5842 could be on its limit swingwise with 150V at the plate.





See that you get the amplification you need WITHOUT cathode bypass caps !!





Dunno if the 6B4G is an indirectly heated tube; in any case i would go with directly heated tubes for the output stage. I hear the Sovtek 2A3 monoplate is available in premium matched pairs for comparatively low money and it sonically outperforms NOS RCA 2A3 dual black plate, moreover it stands more dissipation at the plate, about 18Watts, and higher plate voltage. But, please, buy them from the official distributor, not from dubious sources. A lot of 2nd and 3rd selection grade out there....





I second Mohan's comments. He makes hints minimizing the bypass caps influence. Better get rid of them :)


PS ripple filterning is best done using a choke input, i totally agree.
See Radiotron Desinger's Handbook, 4th ed., pp 1162-1163.
But even fast soft recovery SS diodes produce way more HF (and also spit way more HF back into the wall plug, disturbing other audio gear) than vacuum diodes do. Experience shows that real life chokes do not completely block the HF form SS rectifiers. Resonances, stray capacities, etc., recently got a lecture about that from my friend Manfred myself, just hand it to you :) .

I would go with tube rectifiers; but this is quite a decision as you will need another mains transformer; voltage drop is higer and not as predictable and ususally you will use a centertapped secondary supplying 2x the voltage you need (r would you use a Graetz bridge rectifier?). From sonics, this is no question to me, tube rectifiers. Once you had them properly working, you'll never will want to miss them again.


Recommendation from the Munich Triode Mafia: use TV damper diodes as PS rectifiers such as 6AX4, if one pair is not enough, parallel them or use stronger types. They are dirtcheap and usually do their job better than double diodes like 5U4. U need 2, but they are smaller, stand higher voltages and currents, have better switch-off characteristic.





ECC99: This tube is a new development by JJ and rumours i heard from other users tell nothing but raves. To what i hear, the JJ ECC99 is vastly preferred to other choices like 5687, 7042, E182CC in Lynn Olson's Raven preamp sonically.

Methinks this is a very good choice for the input stage and as it is a double triode, you can use it PP and with input transformer as i suggested above.



As soon as my amp is running, i will try the ECC99 out myself, but against a strong competitor: i own a few pairs of the 6900 which is "the military version of the military version" of the 5687 and reputed to be the sonically best small double triode; from 1st listening tests I can second that, the tube sounds better than the 5687 to an extent utterly unusual for comparing different brands of the same tube.

But, the 6900 is muchodinero :( whereas the ECC99 is cheapocheapo.



BTW, the 5687 is not bad at all, it is a really good sounding tube.

...The LL1635 is replaced by a LL1660. ...


Why? the LL1635 is a worldclass performer as 1:1 IS trannie.

Other turns ratios than 1:1 tend to be a bit problematic for poweramp IS use (look at the bandwidth limitation). I plan to use the LL1660 as my preamps output transformer (have it already) and plan to use the LL1635 as interstage trannie in my PP power amp.



So if you have an LL1635 PP or LL1635 5mA you do not need and consider, i would love to buy it :) otherwise i have to order it from Lundahl.
 
Wow!! thanks for all the input!

Firstly, let me state that I consider this project to be my first serious attempt to build a tube power amp. (though I have several SS amps behind me).

|Any reason for your preference to ECC99?|

The ECC99 have gotten lots of attention lately, and being a cheap & readily available tube with great curves I really really like it. The reason NOT to use the (great, might I add..) 5842 is because I already have this tube in my audio chain, in my Phono pre-amp.

Hmmm, think I'll try out -40-45 volts then, giving a cathode current of ~45mA/tube...

The 6B4G is a DHT, but it is new to me that the Sovtek 2A3 can handle ~18W... interesting... should be theoretically better to use a 2,5V winding than a 6,3V one. Fewer turns for the signal to pass through... but this is a bit far out perhaps...?

|There is no reason for not using a PP driver stage and an input transformer (except the almost unhearable protests of your wallet) but sound will become even more glorious with 3 trannies in the signal and the driver can be made more juicy. The 417A/5842 could be on its limit swingwise with 150V at the plate. |

(just like in the Lynn Olson Amithy / Aurora amps?)
As stated above, this will be my first tube amp. I like to keep it not-so expensive, and my described solution is pretty much the simplest you can get. As for the BP caps, they will be the experimental part... with or without... But I really see your point! :)

|...use TV damper diodes as PS rectifiers...|

The 6D22S is a new TV-damper Diode from Svetlana, featuring 30sec warmup time and ultra low noise!

|The LL1635 is a worldclass performer as 1:1 IS trannie...|

Well, I plan to use the IT for SE-to-PP conversion, and Lundahl web-page stated that LL1660 was preferred with this kind of work... Dunno what reasons though... and the LL1660 being quite flexible, I can use it widely in other circuits when (not if...) I want to try new amplifier circuits. :) The LL1660/10mA will be used in the 1:1.125 configuration.
Though, the LL1660's with the ECC99 will leave a low freq. cut-off at 11 Hz (-3dB)... a bit high, no? With the LL1635 even higher...

And, no, I have unfortunately not got any IT's yet... But I DO have a pair of TANGO FW-50-5 OT's that I'm goin to use!! Got them used for a good price! Oversized for this amp, but gorgeous beautiful irons!

|BTW, the 5687 is not bad at all, it is a really good sounding tube.|

Yeah, I guess this family of tubes are all winners. Interesting about the 6900! Heater pulls 1 amp! Have never seen anyone sell thisone though... :(


--------------
Stig Larsen
 
Stig,

Just a little Note for what it is worth:

All (well most of the ones that I have had patience to try) tubes have narrow windows or select/preferred set of operating conditions within which they perform exceedingly well. This is not directly related to their power output or dissipation ratings.

For example a 300B operating at Va=250, Ia=30-35mA is in musical terms more satisfying than its operation at higher power levels Va=400V, Ia=70mA. I would not call this “Tweaking” but would call it “Knowing your tube”. My experiences are using Lowthers and they need only 60mW to get them going! Once you have your amp running, I am sure, you will optimise the operating conditions to suit the sound quality that you prefer. So leave a little headroom in your power supply designs.

There is no substitute to DC coupling. Since you have elected to go for PP topology, this is not a practical idea.

Constant current regulators in heater supplies (separate supplies for each tube) would not hurt either. You can also vary the heater voltages to control the emission characteristics and thereby the Ia-Va curves. Eventually, you will end up believing in Relativity.

Mohan
 
Yeah I guess thats a good idea to have some headroom. Maybe fixed bias on th output tubes? Easy to implement through the IT, and would be very easy for me to do some finale adjustments just using my ears or something... Would need an extra winding though.

CCS: do you mean for the output tubes? Never seen that. I've only heard talk of using CCS's with DHT's on DC....(in SE)

I've seen some PP DHT amps that use one heaterwinding for each output tube, and then connecting the CT of the winding to a common cathode resistor. Whats the score here?

|There is no substitute to DC coupling.|

Hmmm, wouldnt this call for the need of a BP cap on the cathode? (with choke loding...?)


Stig Larsen
 

J Epstein

Member
2002-02-08 7:24 pm
I've seen some PP DHT amps that use one heaterwinding for each output tube, and then connecting the CT of the winding to a common cathode resistor. Whats the score here?

Probably the transformer selected had two filament windings, each with only enough current capacity for one tube. You can connect these two windings in parallel and run both tubes off this composite, or use them seperately if you want. I don't see why it should matter. If you do parallel the windings, I think you do not want to use BOTH center taps, just one, or else the two windings may burn extra current "fighting" over a slight mismatch between them.
 
Stig,

Last paragraph in my previous post is a general comment for DHT applications. I should not have used the word “Constant”. A current source limits the inrush current to cold heater and thereby prolongs tube life. Yes, I recommend it for Single Ended designs.

Yes, there is no substitute to DC coupling and Yes bypass caps ruin the quality of sound output. Unfortunately, we have to have a compromise and I believe in minimalism. In a DC coupled design, the amplified signal is delivered to the following stage in a manner that could be described as “Less modified “ than with other options using coupling capacitor and/or interstage transformers. I simply minimise the “evil” i.e. the value of cathode bypass caps. Without the bypass cap, the tube’s operating point wanders with the input signal.

By the way, I have a personal prejudice towards Directly Heated Tubes in general. I look at the cathode as a series of resistors that oppose the signal current through the tube. The end result is compression, though not as much as that is applied in some recording studios! Sometimes, I wonder as to how much of this signal energy is used up in heating the filament!!!

Mohan
 
hey...

Well, haven't started building yet, still collecting parts. I have re-designed the schematic quite a bit, as can be seen below.
Any comments?

[IMGDEAD]http://www.reisa.no/images/6b4g.jpg[/IMGDEAD]


I have completely removed the BP-cap at the input. I have finally understood that there is no need for it.
The PowerTranformer is a custom-wound toroid and oversized quite a bit...

And yes, I will run dual mono!

Is it a matter of getting the PSU caps as small as possible too? In SS equip., it's always a matter of having as many farads as possible.

Tube amps rarely use anything bigger than 220uF...

-stig
 

yves

Member
2001-02-06 6:18 am
earth
Hello.

Great thread, and great people:)

Stig, you may want to try the ultrapath
connection from IS psu end to the cathode
of the first tube. You can see the
circuit in Electra-print transformer
website. It is a simple matter to
move the psu cap from B+ ground to
cathode. I think there is no harm
at all done.

Oh, and where can I see comments of the
ECC99 (positive or otherwise) on the net?

rgds
Yv
 
CCS: do you mean for the output tubes? Never seen that. I've only heard talk of using CCS's with DHT's on DC....(in SE)






Stig,





exactly this is what i am planning with my PP AD1 amp.
:)

Hint: Have two power tubes and a CCS forming a differntial pair, a "long-tailed pair". If your CCS does't make the voltage across it alone, put it in series with a resistor biggenuff to dissipate the power.





Kirchhoff's law #1 applies here: both cathodes are linked to the same node: all current going into a node and all currents goiing out of that node sum up to total zero current. This means that the currents going into both cathodes are equal to the current fed into the node by the CCS which is constant !! (assuming real electron flow, not technical current direction). Accordingly both output tubes have to act exactly inverse instead of sloppily so as usual in common PP amps with both cathodes referenced to ground separately.



I heard an amp having this feature, it combines SE and PP virtues, best of both worlds.
 
|you may want to try the ultrapath connection from IS psu end to the cathode of the first tube|

I think maybe the 0C3's would have a problem with capacitors like this in the load, but I might be wrong...
Is the ultrapath like this supposed to shorten the signal path to ground or is it to reduce PSU ripple? Should the value of this cap be as if it where a normal BP-cap? Could it be smaller than that?
This cap would have to be of high quality, yes?

Don't know of any "offical" test/reports on the ECC99, just have read about it here and there... try a goggle search.

Dice45;

I'm sorry, but I don't think I got it strait... But it sound interesting! (your amp too!) Got any schematics?
 
Stig,

Choke input filter is great. My personal preference is for a two stage LC filter(5Hy-100uF followed by 5Hy-470uF). Signal to noise ratio should then be around 68-70dB at full power output. I use Panasonic (marked UP 450V) caps.

Yes, if you wish to have very good transient response, then you could consider large value chokes and film foil (not metallised) capacitors. Because it is Class A-PP, we cannot expect significant changes in current demand and so a reservoir may not be required.

Perhaps you could consider dropping some of the B+ voltage in case the maximum allowable Va for ECC99 is limited to 200V (I do not have the specs for this tube). I am referring to the VR tubes and the 5.1K Caddock resistor. I would rather add an RC filter here and use the R (max 1K; the lower the better) to drop the voltage as required.

Input tube is biased around 6V. Is this correct?

Mohan
 
Hmmm, yeah. I just used one LC stage as in the original schematic because obviously this is enough to reduce ripple to acceptible levels. And with the Ultrapath connection on the output tubes, hum will be even lower.
Does anyone know HOW MUCH a PP amp "reduce", or "cancels out" ripple at it own?

As for the ECC99; http://www.jj-electronic.sk/tubes.htm <--- theres the specs. 400V, 60mA, 5W in short. Well within ratings. (~4 watts)
The 5.1k resistors are paralelled. If to much ripple I will replace these resistors by a 1k and a 1,5k in series with a 47uF cap in the middle.

6V is the correct bias voltage. I like to keep the bias somewhat higher than normal. 2V RMS (from a CD player etc...) is in fact around 4-5 volts, no? And transients may also reach pretty high levels I recon...
Don't know if this has any worth doing though, but it feels right. :)

Stig
 
Originally posted by stigla

I'm sorry, but I don't think I got it strait... But it sound interesting! (your amp too!) Got any schematics?




Stig,

before i post a schematic not tried out and running stably, some wraith shall eat my soul!

I just try to explain how a differntial amplifier stage works. Have a look at
Allen Wright's PP1 amp : the two 1000µF electros shorten the cathodes AC-wise which means that for AC, both cathodes are connected by the 68Ohm resistor to ground. Resistors have a tendency to keep currents constant, not as much as an inductor would do, but the bigger the resistor value is, the more it tends to constant the current (470k looking to minus 500V would do a might fine job here).
How bad the two caps acting as two coupling caps in series are electrolytics! (and have to be)

Now imagine the EL34 not having this 470Ohm || 1000µF in the cathode but their cathodes being tied together and connected to ground via a constant current source. As is done with the lower cathodes of the 1st stage's differential pair of cascodes.
The CCS keeps the total current through both EL34 constant to frequencies way above the audible spectrum. So if on EL34 reduces its current, the other one has to increase its current exactly to make the total fit (the CCS will see to it :( ) . Result: both output tubes have to act exactly inverse. Any sloppiness is prevented by the CCS, be it DC or AC. It would be interesting how this works if th CCS is replaced by a choke which only works as a constant current device from a few Hz up and so biasing would be easier.

Biasing, good i mention it, you have to add additional bias to make both tubes have the same quiescent current and you have to check and re-adjust that regularly. Otherwise the OPT goes into saturation and you have a mess instead of good sonics.

To my amp: i am trying something very similar, but my design is spiced with danger, i decided to settle on a differntial pair of 71A driving a differntial pair of AD1, both differntial pairs having a CSS at the cathodes. The spicing danger: both 71A and AD1 are directly heated tubes and i may face more noise picked up by the virtual grounds (common cathodes) than i may find bearable. Getting the noise refencenced to ground out of the heater supplies will be really tricky. This doesn't work with line levels but it may work with signal levels and S/N ratios in a power amplifier, any stage having a gain of 3 or so.The 1st stage too is a differntial pair, it uses the indirectly heated 6900 which are choke-loaded i.e the plate load is a choke. Between 71A and AD1 i use an interstage trannie: Lundahl LL1635 .
 
Hmmmm....


I've seen at Tube CAD some ideas using an inductor at the cathode to bias a common-cathode pre-amp. But no indications or hints to how to calculate the value of such a choke, but the value may be un-critical...
I liked the Idea, and I even considered building such a linestage using the '99 and two LL1668, but I changed my mind.

So the only problem then, with CCS's are biasing of the tubes? Fixed bias comes to mind... Why the regular bias check? Because of the drifting of the SS devices in the CCS?

How critical is the coloration of the CCS?




stig