• 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.

Kofi Annan in: "Tube Amp for Multi-Way Speakers"

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The EICO HF-87 was basically a stereo version of the Mullard 5-20 but with a 6SN7 for the LTP and a half a 12AX7 as the input/voltage amp.

If the 18k 1W tail resistor (from joined cathodes of 6SN7GTB to ground) is changed to a constant current sink (CCS), the plate resistors of the 6SN7 can both be 33k ohms (instead of one being 33k and the other 28.75k).

I'm sure there are lots of other tweaks possible too.
--
 

Attachments

  • eicohf87.gif
    eicohf87.gif
    17.6 KB · Views: 372
Administrator
Joined 2004
Paid Member
I still think 45's suggestion of paralleled El84 has merit, very easy to drive and at least 30W in PPP pentode.

I don't understand the negative comments about current hogging, individual auto-bias is easily arranged and will prevent serious problems, yeah they won't match perfectly but I doubt that is a major concern in this application. I have built some PPP amps in the past for clients and they're all still running today +20 years later. A little bit of thought and the problem is easily solved.

You can drive a quartet of them with split load (cathodyne) comprising a 12AT7 shared between channels and a single 12AX7A also shared for the first stage voltage gain.

Full wave bridge or doubler, single choke, transformer with a single HV winding and couple of filament windings and you are good to go.

Simple, inexpensive and with a little thought will work well. 10 inexpensive tubes for 30Wpc stereo, not bad.

Or horrors do the same thing with a pair of 6L6GC per channel for 6...
 
Because of current hogging, forget PPP EL84/6BQ5s in a self biased amp.

That's really trivial to sort out: back bias with individual pot in the worst case. So no need for additional secondaries.

Anyway I didn't have any problem and used one pot for each side of the PP. Having 8 tubes is easier to match them.

I used the Silvanya 10GK6 which is 10v version of 6GK6, an up-rated EL84 (different pinout which allows for 330V plate and G2 voltage). This is still available in good numbers at a fraction of the price (I paid $3 each). Better quality than current production JJ et similar....

Any designs you would recommend?

The amp I built with the 10GK6 was my personal take on existing amp designed and published on Italian review in the late 90ies. I have that number and could scan the circuit. The output stage is PPP of EL84's running in (almost) class A for 20W power. Back bias from the main supply and Vg2 regulated by means of mosfet + jfet as error amp.

I added the cathode feedback and run in class AB.

Here the bias can be tweaked in 2-3 way depending on what one wants. Can be as the original (i.s. only passive RC filtering) which will cause some bias shift like in cathode -biased circuit (not a bad choice a priori if one gets a soft clipping) or regulated. You might even start with a rough regulation using zeners and then improve that once you get it going. Regulation does not need to be complicated. Have a look to the Red Light District amplifier in this forum for example.

From the design point of view I only changed the front-end. I used a board that I already had in my box consisting of a Williamson style front-end with 6CG7's.

In general what do you need to change? Once you have the cathode feedback at 10% it just means that 10% of the output voltage at the primary of the output transformer comes from the cathodes of the EL84's and so, because there is no gain there (actually a bit less than 1), it's the driver that has to deliver it.

30W into 4K means that each side of the PP is outputting 245V peak in total. 10% of that is 24.5V. Make 25V peak. Add to this the normal swing for the plate output 12-13V peak and that makes no more than 40V peak. Basically any frontend that can drive EL34's properly will do.

P.S.
What the difference respect to using more standard ground fbk with EL34's?

The difference is that, if the transformer is good, you don't need to add other fbk and amp is easier to make stable but more importantly the reduction in distortion is more effective for the same amount of fbk.

For the PPP with EL84's into 4K for 30W and 10% cathode fbk the actual fbk in dB's is just under 10 db but the reduction in distortion is about 19dB!
You don't get that with loop feedback.

The reason for this is that you first make the tubes linear and then combine them. They really behave as very linear triodes with the efficiency of pentodes.
 
Kevin / 45,

Thanks for responding!

45-- I'll sketch something up with your recommendations in mind. I've never implemented cathode feedback before. Not sure that I can get a transformer with cathode feedback windings, so wouldn't I have to tie the feedback loop to an unused tap on the transformer secondary? Is that how you implemented yours?

Kofi
 
I put a link to a good and reasonably priced transformer in my first post. It's well worth the price.

You can use the secondary for CFB but usually the amount of fb is less and the secondary has to be such that you can get a symmetric configuration.

I normally design and get done the transformers (especially in cases like this) but also have commercial products.

Regarding the tubes...you can find those 6GK6, 10GK6 and 16GK6 at Vacuumtubesinc at moderate price. They have got plenty so you can buy 2-3 times the amount for the same price of one set of properly selected EL84's and give your friend 1 or 2 sets for replacements. They will outlast him I guess...:D

The 6GK6 also has 13.2W max plate dissipation rating vs 12W of the normal EL84. It's a bit like the Russian military 6P14P-ER but different pinout.
 
Maybe try the EL86/6CW5 if going for EL84 types? 20 watts is easy, and they need less impedance than the EL84 does (3-5K a-a is good) making a nice transformer easy to design/source. They sound great, same pinout, and look the same as EL84. They are cheap too.


Also, we will be expecting airplane or at least helicopter rides for the advice, 'tis only fair :)
 
Hey! Sorry I vanished for minute. My kid actually spoke to me for about ten seconds and I couldn't get over it...

I have attached an LT Spice sketch of the EL84PPP using a 12AX7A into a 12AT7 cathodyne as Kevin suggested. Open loop it models about 50W peak output. Not bad.

In LT Spice it's running into a 5K primary. Once I fugure out how to lower the model to K and allow for cathode feedback we should get close to reality (note: LT Spice really does not like modeling NFB. Not. One. Bit. Acts like it's calculating PI to the last digit while developing cold fusion).

Any values look incorrect here?

I'll respond regarding the other posts in a bit-- just wanted to get this one out there for review.

Kofi
 

Attachments

  • EL84PPP v1.PNG
    EL84PPP v1.PNG
    185.4 KB · Views: 322
That looks lovely. My first impression is that C5, 6, and 7 are way bigger than needed. The biasing scheme for the concertina is interesting, any reason for this over self bias?

Yep. I used the fixed bias because I was initially unable to get LT Spice to deliver a uniform phase split from the self-biased cathodyne on account of I blew it and forgot about the bootstrapping resistor.

I reconfigured for self bias and reduced the coupling caps to .33uF on all stages (see below). Will that suffice?

More to follow...

Kofi...

Oh, and would a CCS in the voltage gain stage help here? It would be cool to have CCS on the power tubes, but I really don't wanna make EIGHT little CCSSEESSESS.

Kofi
 

Attachments

  • EL84PPP v2 - Self Bias Cathodyne.PNG
    EL84PPP v2 - Self Bias Cathodyne.PNG
    183.4 KB · Views: 297
Administrator
Joined 2004
Paid Member
Definitely moving in the right direction, I would reduce those 0.33uF coupling caps to about 0.1uF to the grids of the EL84.

I don't think C1 - C4 need to be larger than 47uF - 100uF each.

I believe there is sufficient gain that when you close the loop you will not need more than what you already have - therefore the CCS is really not necessary.

Simplicity and components unlikely to fail are what you want here.

This is a lot like the original Eico HF30 design (had a pair long ago) and should be very reliable.
 
You're doing a lot of things right here- and the above suggestions are on point. I would use anything from 100-470uF for C1-C4, and I would go for something with a 105C, 10,000hr rating, They'll last a good long while, and at the smaller values are still very affordable from places like digiKey for good name brand commercial parts. C4, C5, and C6 can all be 100nF. Don't bother with audiophile secret-sauce parts here.

Individual cathode bias for each EL84 is a lovely touch, practical and reliable :)

For the CCS, well, that's up to you. I would go for a plate resistor from 120k to 150k and leave it at that. I happen to find 12AX7 types do well without a CCS so long as the cathode resistor is unbypassed, and the plate load is high. Play with the parts until you get 60~70% of the B+ voltage at the anode. Gain should be 50~ or so at the 12AX7 before feedback. Plenty for EL84 types in this application.


This is going to be one sweet amplifier. :cool:
 
OK-- lowered the cap values (see attached schematic). I believe increasing the 12AX7A plate resistor would require an increase in B+, which I think would cause some issues (unless I am misunderstanding, which DOES HAPPEN), so I'm inclined to leave it as is.

Also, it looks like I will need to have power transformer that is capable of delivering about 260mA of current at around 275 volts. That seems like a tall order (unless I am misunderstanding which, please refer to the PREVIOUS SENTENCE).

Thoughts on the power transformer? Also, I'm assuming I should go with the Toriody transformers with the CFB taps? Any thoughts on that?

KOFI???
 

Attachments

  • EL84PPP v2 - Self Bias Cathodyne v2.PNG
    EL84PPP v2 - Self Bias Cathodyne v2.PNG
    163.9 KB · Views: 301
Coupla things that I hope will help...

I think R12 and R13 are not in their correct positions. R13 (grid leak) should go from input to (signal) ground before the connection to R12 (grid stopper). See attached schematic.

The way you have it now, R12 and R13 form a voltage divider, albeit one that doesn't knock down the signal voltage by very much.

Also, I noticed that R14, the cathode bias resistor for the 12AT7, is a pretty high value, keeping the plate current of the 12AT7 to less than 2.5mA. Is that enough current sunk to drive the Cmiller of the paralleled EL84s in UL configuration? Would Ia of 3mA or 3.5mA be an improvement, or not necessary?

I believe increasing the 12AX7A plate resistor would require an increase in B+, which I think would cause some issues (unless I am misunderstanding, which DOES HAPPEN), so I'm inclined to leave it as is.

The way the 12AX7A is configured now, it will be drawing a healthy (for a 12AX7) 1.2mA of plate current. 12AX7 can work well with less than 1mA of plate current. What would be the harm of increasing the value of R21 (the 12AX7A plate resistor) to 150k? Or you could re-bias that stage with maybe R12 = 220k and R22 = 2.2k. LTspice says that would put 168V on the plate, 1.5V on the cathode, plate current of 0.69mA. You'd get a bit more gain out of the 12AX7 that way, but transconductance would be a bit lower. Would that lowered transconductance cause problems driving a negative feedback loop? It certainly should not be a problem driving the 12AT7 cathodyne (that will be easy to drive, high input impedance, low input capacitance).
--
 

Attachments

  • _KOFI_12AX7-12AT7-cathodyne.png
    _KOFI_12AX7-12AT7-cathodyne.png
    17 KB · Views: 267
Last edited:
Coupla things that I hope will help...

I think R12 and R13 are not in their correct positions. R13 (grid leak) should go from input to (signal) ground before the connection to R12 (grid stopper). See attached schematic.

The way you have it now, R12 and R13 form a voltage divider, albeit one that doesn't knock down the signal voltage by very much.

Correct! Thanks for catching that!

Also, I noticed that R14, the cathode bias resistor for the 12AT7, is a pretty high value, keeping the plate current of the 12AT7 to less than 2.5mA. Is that enough current sunk to drive the Cmiller of the paralleled EL84s in UL configuration? Would Ia of 3mA or 3.5mA be an improvement, or not necessary?

The way the 12AX7A is configured now, it will be drawing a healthy (for a 12AX7) 1.2mA of plate current. 12AX7 can work well with less than 1mA of plate current. What would be the harm of increasing the value of R21 (the 12AX7A plate resistor) to 150k? Or you could re-bias that stage with maybe R12 = 220k and R22 = 2.2k. LTspice says that would put 168V on the plate, 1.5V on the cathode, plate current of 0.69mA. You'd get a bit more gain out of the 12AX7 that way, but transconductance would be a bit lower. Would that lowered transconductance cause problems driving a negative feedback loop? It certainly should not be a problem driving the 12AT7 cathodyne (that will be easy to drive, high input impedance, low input capacitance).
--

I upped the plate resistor for the 12AX7A and lowered the bias resistor for the splitter. I also ATTEMPTED to model a cathode feedback transformer, but things look very wonky to me here.

Did I get the CFB arrangement and modeling correct? Also, regarding the power transformer, am I correct that for a stereo supply I would need over 360mA?

Kofi

PS-- Credit where credit is due: I modeled the CFB using the design in this thread.

Did I get the CFB arrangement and modeling correct? Also, regarding the power transformer, am I correct that for a stereo supply I would need over 360mA?

Wait-- should be half-duty cycle, so I should only plan for 180mA, right?

Other Kofi
 

Attachments

  • EL84PPP v2 - CFB v3.PNG
    EL84PPP v2 - CFB v3.PNG
    171.8 KB · Views: 237
Hmmm... That way of modeling the cathode feedback secondary doesn't look right to me. I have to think about how to do that in LTspice. I don't think I've ever tried.

Does your OPT have an extra secondary winding for cathode feedback? If so, it would be wired like this:
n9yox.jpg


or this:
Fisher_100_schem.gif


I would first design (and model) the amp without negative feedback, and get the best looking circuit you can get before adding NFB. Then once you're satisfied you've done the best you can within the limitations of devices and available voltage and current from your supplies, you can start adding NFB knowing that the underlying circuit will give you the best chance to get optimal results.

Re: power transformer - It's best to have enough current capacity from the transformer to be able to supply the peak current demands the amplifier will draw from it.

Are you figuring each EL84 will draw 45mA plate current with no signal? Then yes, your power transformer has to be rated to deliver at minimum 360mA (45*8), plus the current required from the (class A) driver stages. That's 2.5mA for the two 12AX7 triodes and 6mA for the two 12AT7 triodes, so another 8.5mA on top of the 360mA for the EL84s, or 368.5mA. I'd call it 370mA at minimum. I'd go for 400mA current capacity. That's going to be a pretty hefty chunk of iron.

In VA terms, that would be 250VAC * 0.4A current = 100VA.

Antek makes a toroid with that rating for under $40 -
AS-1T250 - 100VA 250V Transformer - AnTek Products Corp

Edcor makes a 500VCT 400mA EI core monster too (9.75 lbs!), for $75 -
EDCOR - XPWR078

Those are in the USA, so you might need to look closer to home.
--
 
Last edited:
The Antek AS-1T250 says that it's rated 100VA (250V 0.4A). It has two 250VAC secondaries (not center tapped). For a stereo amp, I guess you could put a full wave bridge on each of the two 250V secondaries. So, a question: Does that mean each of the two 250V secondaries is capable of delivering 200mA?

If yes, that means the Antek toroid gives you the same voltage and current capability as the Edcor XPWR078 EI cored 500VCT 400mA transformer at less than half the weight. (9.75 lbs for the Edcor vs. 4.1 lbs for the Antek toroid.) And...

The cost for the Antek is something like half that of the Edcor.

Really? If so, looks like Antek all the way.
--
 
The Antek AS-1T250 says that it's rated 100VA (250V 0.4A). It has two 250VAC secondaries (not center tapped). For a stereo amp, I guess you could put a full wave bridge on each of the two 250V secondaries. So, a question: Does that mean each of the two 250V secondaries is capable of delivering 200mA?

If yes, that means the Antek toroid gives you the same voltage and current capability as the Edcor XPWR078 EI cored 500VCT 400mA transformer at less than half the weight. (9.75 lbs for the Edcor vs. 4.1 lbs for the Antek toroid.) And...

The cost for the Antek is something like half that of the Edcor.

Really? If so, looks like Antek all the way.
--

That would be awesome! I was looking at chokes last night, however, and it appears that I will need to split the supply out with an individual choke for each channel. The current ratings of chokes are too low unless you go HUGE and I'd rather have the chokes smaller and 'under the hood'.

Make sense?

Kofi
 
Yes. Since this looks like it will be a BIG stereo amp chassis, and you'll have individual plate supply windings from that Antek transformer, you might as well go dual mono with the power supply design. It will be more expensive that way, but oh well.

What chokes were you looking at? There are some decent ones available for not much money.

The Hammond 157R looks usable, 2H with 57 ohms DCR at less than $25 each.
If you need cheeeeeap, there's the Triad Magnetics C-24X, 1.5H 50 ohms for less than $10 each.

I don't think you'll need large inductance values since you'll be using a cap input power supply (as opposed to a choke input supply), and 400V rated capacitors will be fine for a 325V B+. There are good 330uF 400V caps available. Perhaps use one of those for the cap after the choke, with a 100uF 400V low ESR/high ripple current rated part for the reservoir capacitor (before the choke).

Or did you want to go more deluxe than that?
--

On further thought...

To arrive at the voltage to expect for the B+, you'd have to factor in DC resistances in series with the power supply. So...

Let's say the filter choke's coil resistance is 50 ohms, and the primary winding resistance of the OPT is 100 ohms. Given that, then...

with 170mA current drawn from each channel, 150R * 0.17A = 25.5V dropped across the winding resistances.

Looking at that Antek transformer's spec sheet, I see that under load, if 400mA is drawn then the secondary voltage will be 241VAC.

AS_1T250__62569.1387573726.1280.1280.jpg


241 * 1.35 (conservative estimate of rectifier efficiency) = approx 325VDC

That means the B+ to the EL84s might be as low as +300V in the end.

The next step up the Antek 100VA line is the AS-1T275 with nominal 275VAC secondaries. That one shows 264VAC delivered under 350mA load. That would give you a higher raw B+ voltage, something like 355VDC. The 50 ohm DCR of the filter choke would drop about 9V (50R * 0.17A = 8.5V) for approximately +335VDC, then another 17 or so volts dropped across the OPT primary winding resistance, so perhaps +330V delivered to the EL84 plates.

I figure it's easier to drop a few volts than it is to add some. So I'd look at the 275VAC 100VA transformer, Antek AS-1T275, instead of the 250V one. If the B+ voltage comes out too high, you can always put a small value/high wattage wirewound resistor in series after the rectifiers, just before the reservoir cap (power supply input cap). That is sometimes necessary to damp subsonic resonances anyway. 22 ohms would drop almost 4V, for instance, or 47 ohms would drop 8V.
 
I was just modeling this in PSUD last night and came to a similar conclusion regarding the voltage drop under load. I modeled it at 290VAC and found it came in around ~340VDC. I'll give the 275VAC job a try...

Here's my quick sketchup of the EL84PPP with EI transformers. I'll work up a couple with the toroid power transformer as well.

I believe this is the toroidal output transformer that 45 mentioned. It has cathode feedback taps at 10% and can handle 300mA. Just wondering about DC balance, though-- I think there's a risk of core saturation unless we can ensure balance. Any experience with this?

Thanks so much for all the responses. I could not do this without you guys.

Kofi
 

Attachments

  • Screen Shot 2019-08-18 at 4.36.26 PM.jpg
    Screen Shot 2019-08-18 at 4.36.26 PM.jpg
    264.5 KB · Views: 87
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.