Designs on your noob, err, tube

[PeteN]

Hello everyone,

This is my first post, so hope I don't get anyone peeved. Trust me, I did search (not neccesarily here, in every case), so regarding any idiocy I demonstrate below, please be gentle. There seems to be a dearth of information in the area between 'what's a valve?' and 'How to configure the quad-quasi-quantum-octode-megaquench-converter for construction of an RF cookie baking oven (volume XXXVII)' (i.e, really high level stuff), or the sites that explain something don't do so in a way that a total idiot [self] can understand.
(What's that? Total idiots shouldn't play with valves? Why, that's descrimination! I have every right to poke my finger where ever I damn well-- zzzaap. Ow.)

Anywayyyyyy.
This post will be long, and covers several points and questions so please bear with me here, I didn't want to spam the thread listing with seperate questions within my first week, you know?

-x-
(Quick links: Schematic here, PSU Schematic here, psu graph here)

So a month or so ago a friend of mine announces he bought a tranny amp off eBay, I start to look around on the net, and next thing you know, I find myself pondering the fatal question "I wonder if I could scratch build a valve amp?" (creating my own valves and everything... hehe but seriously.)
Initially I knew almost diddly squat about valves (wait...sorry... tubes), but I've been interested in electronics for most of my life.

Now, why design one when there's plenty of decent kits/schematics like the bottlehead amps and the 'classic' Mullard 5-20? Well, that's just me. Once I get an idea into my head, if I like it, I just can't get rid of it and that's what happened here. It started out as (basically) a copy of a UL PP amp from another website, but then I changed things gradually as I read more, and ended up with this... this... thing.
I don't know if I'll actually build this amp, it's going to be quite expensive even with cheap transformers, but the mental exercise was interesting at least, and I learned quite a lot, and would like to learn more. Plus there's the good chance it wouldn't work [very well] in the first place!

Long story short I started researching and poking around the web and several good articles/sites later* I settle on a P.P. using two EL34's (available to me). This choice was partly because I'm aiming for around 25 watts (per channel), and with ~12.5W max output power per valve, they seem perfect. Furthermore I've read around a bit and came across this thing called 'Ultra Linear' which, upon investigation, seemed to involve wiring one of the grids in each triode to an additional tap on the transformer (optionally through a small resistor), and this making the whole thing magically sound better (before I get lynched, ok, sound different) - which from my understanding is same power, less distortion. Perfect! What could be simpler?
(* The sticky at the top of this forum is super, but would be friendlier to my connection if it could all be condensed from a zillion pages into the first post!)

Next major issue was driving the EL34s from 'line level' - PC soundcard/mp3 player/dvd player/whatever - which I understand to be 'about' 1V, but we'll get to that.

I searched a bit more, and wearily came across the Boozehound site (and that's how I got here, actually), and perked up when I found his 'Tube HiFi howto'. I've followed this through once or twice, applying the relevant bits of the math/methodology to my design-to-be. From this how-to, and from the datasheet for EL34, I settled on an operating point for the valves of U_a = 250V, I_a = 100mA, U_g = 12.5V, into a load (OPT) of 5k5 Ohm - from looking around that seems to be about the average.
From the operating point I had, I went and abused Ohm's Law a little, and established that I need a resistor of 125 Ohms between Cathode and Ground. I picked 250V as it seems to be about the average for this kind of "low" power amp, and the transformers are affordable.

From my understanding, now I have an operating point with R_gk of 12.5V, I need a signal that is in the range -12.5V to 0V to drive the EL34 from cutoff to saturation.
Next was the problem of getting from a 'line level' signal down to this range. Well, from research it would seem a dual triode is a common way of doing this. ECC82 (12AU7) is one such dual triode. Gain of around 15, so looking good. I perused the datasheet excitedly. I looked at the 'typical operation' table and saw, to my delight, that 250V and ECC82 can coexist peacefully, yay!

Anyway I perused the datasheet more, and to my continued delight I came across a typical configuration for a phase inverter, of the form 'concertina'. Brilliant! So I take a look, and [try to] figure out how it's working. First triode is amplifying [dunno how to figure the gain out though], second has a gain of 1 as the resistors R_{a, k} are both the same. We take V_out from the anode, and V_out' from the cathode. Both go through a decoupling capacitor this came out as 0.01uF and a 680k Ohm resistor between there and ground, for both.
The 680k resistors would appear to be serving also as the grid resistors (to reference them to ground) of the EL34s as well. This was in accordance with the way the boozehound site did the driver stage too, huzzah!
Looking down at the little data table below the example schematic, I find to my fortune a coulumn headed with my chosen 250V! As if by magic, gain appears to be 11 for this particular configuration (I needed (-)12.5V, and input signal will be "about" 1V, so a gain of 11 should be quite sufficient).

So, the story so far. I read a few websites, download a few datasheets, and (aside from picking an operating point [no, wait, that too...]) all the work is done for me.
Now we have an ECC82 amplifying and phase-inverting the input signal to about 11x, which is then fed into the EL34s through the decoupling caps.

Next concern would be the output transformer... That's a complex case of having a lot of money it would seem. Though electically, it seems fairly simple. Valve Anodes to Transformer Anodes, 250V to the centre tap, and UL taps to the screen grid of the EL34s. Small screen resistor optional (seems to exist on about 50% of schematic's I've seen). On the secondary side I wire the 4/8/16 Ohm taps to some form of connector, as well as the Ground tap, which also connects to ground on the power supply. I have descried some cheap (£25/US$50) UL PP output transformers on Maplin (of all sites... they don't even sell valves/holders), designed for no less than 2xEL34s! They're made by OEP (link at end). Opinions?

Finally (amplifier wise), input. Now, this will probably come from 2xRCA cables which will be juiced up by my PC sound card, mp3 player or similar. Firstly, I'm going to guess that the outside terminal of each socket is grounded, and the centre pin will be the source of the input signal itself. This then goes through a potentiometer aka volume control, decoupling cap (prevent DC entering the system? Also high pass of ~16Hz with the 1M resistor?).

So that's the amplifier done (lol). Full schematic at here for your ridicule.

Next issue is the power supply (le gasp!). From reading it seems this can almost be more important than most of the amplifier itself. I've come up with the "Tesco Value PSU" if you like, which is based on a combination of stuff I've seen around the net, plus the fact that I can't afford expensive transformers and chokes (student!).
So... power transformer is a 6N536P (toroidal, but don't run away, we can be friends still). Primary is 2x115V (or 230V), secondaries are 0-280V @ 0.6A and 6.3V @ 7A.
Cost is a 'mere' €68/£47/US$94 (that's a lot for a roll of wire wrapped around a bent iron bar! [joke]).
Rectification wise... well, I'm doing this on the cheap and a 1.5A (Si) Bridge Rectifier costs about £0.20p. Following that, it would seem CRC or CLC is popular. I can afford not CLC, thus I am forced to side with the kinder-to-the-wallet CRC.
Now, I've been playing with PSUD2, and I came up with the design indicated 'ere.
The 56uF cap I can explain; I have an electrolytic one lying around, rated for 400V, infact it'll be nice to actually find a use for one or two of these huge caps I just seem to like recovering from dead things (3300uF, 50V... use, anyone?). Resistor of 1k Ohms to drop some voltage and generally act smooth. Next cap is 22uF - it seems to work in PSUD2, and is a common enough value. Then there's another RC network, again, 47uF is common enough, and the resistor is an arbitrary 400 Ohms which seems like a 'feasible' value to get hold of/make.

At this point (with a 421mA [quiescent] constant current load), voltage available is just over 250V, in PSUD2 ripple looks to be about 0.2V, peak-to-peak. Here's a pretty graph (the wiggles near the bottom are noted merely as a passing curiosity; I think the probable cause is floating-point-whoopsies, though?).

(continued below...)

 

[PeteN]

Part deux

(I said this'd be long...)

One amendment to the schematic; there's now 100 Ohm resistors between the UL taps and the screen grids of the pentodes!

Anyway, now that you're all mostly asleep (yet up to speed!), onto the questions [no order to these, sorry]. Let's start with the noobie ones.

1. Will it work? Have I done something so stupid, so devastatingly moronic that, were I to build this, the universe would simply implode? I hear electrons can't even make up their minds if
   they want to be matter or energy -- so who am I to boss them around?
2. Will it work well? OK so let's assume I have, infact, been going along the right track. Am I going to be greeted with a local AM radio station, white noise, mains hum, or a shrill 'squeeeeee' as one forum member appears to be suffering from, for example?
3. I know this could be better, it's just a case of I don't know how to make it any better. Are there any blatant things I've missed or should have approached from an entirely different angle?
   [*]Grid stoppers: Miller capacitances and reactances and generally anything involving complex numbers I find rather hard to comrehend, outside of a class room. Would it be reasonable to pinch the resistor value from someone elses' EL34 PP and use it here or must it be exact, based on the rest of my circuit?

And you'll really have to work for the marks on these next ones!

1. Output transformers. I can get some 'nothing fancy' models (made by OEP) for £25 each, Maplin code N91CC. They seem like they'll do the job. As I'm no audiophile, would it be worth shelling out any more for the output transformers?
2. Power transformer. I've mentioned the 6N536P made by Amplimo. I can't see any reason not to use this, but I know some folks don't like toroids - again, as a non-audiophile listener, will I think "Ye gad's I'm glad I paid £100 for a replacement made by someone like Hammond"?.
3. Rectifier Diodes. Use a bridge rectifier (4-pin package thingy) or use 4 seperate diodes. As I said 1.5A rectifiers are about 20p, the diodes probably come to the same price so if one is inherently better than the other then this is a performance upgrade I can afford!.
4. Resistor Wattages. This is one of those things that (at least from my searches) just didn't seem to be explained very well, or was never the focus of a page.
   --> I understand that the cathode resistors need to be able to cope with all the power that is going to be going through the valve - does this mean that in the case of the ECC82s, 3 or 4 Watt resistors will do the trick, and in the case of EL34s where one resistor is serving both valves, 25 Watts?
   --> Are there any other resistors in the amp that need to be quite chunky? How about the anode resistors on the triodes, and grid-stoppers for example?
   --> The power supply resistors probably need to be several watts as well. Referring to the PSU schematic, with 421mA load R1 drops ~110V, at 0.11A, so 12W here, R2 drops 22V, 0.1A so 2.5W (3W) here. Is this correct?
   --> For non-high-power resistors, is it ok to use 'normal' resistors, say 0.25W or 0.125W?
5. Capacitors As I've mentioned I'm sort of doing this on a shoe string, so can't afford fancy iron. However, I understand capacitors are crucial too. Your common components supplier like Maplin or Rapid seem to sell HV electrolytic capacitors aplenty, however they're up to 20% tolerance. Would a bog standard 105ºC 400v 47uF fare well in this PSU? I figure 105ºC caps will be more stable w.r.t temperature, which will be handy as things might get quite warm around the PSU?
6. Bypass capacitors. Again there doesn't seem to be a great deal of low-level documentation about how much/why/where. Both stages of this design have them because, in the case of the ECC82, the datasheet said so, and in the case of the EL34s, that seems to be the trend?
7. Volume Pot What calibre... err sorry, resistance, should this be? I've seen schematics with 100k, 470k, 1M and others, but no explanation other than 'this is the impedance the preceeding stage sees'

I have lots more questions but I appreciate I've already taken a lot of time with this already, especially if you more than skimmed over it.

Best regards, thanks in advance for any help/advice that you can give this "toob noob", and once again apologies for the dumb-er of the questions, sometimes I just get mindfreeze and can't figure simple stuff out.

Sites I used:
- Google (search is generally of the form "x (tube|valve) -patent -pdf")
- Boozehounds site (link)
- gabevee's "A push pull tube amp" (link)
- ValveWizard.co.uk
- Loads of others but these are the main ones : )

 

[SY]

For your first four:

1. Yes.

2. No.

3. Sorta.

4. Yes, then no.




I'll stop teasing you in a few minutes. Patience, lad.

 

[SY]

OK, number two. Distortion will be exceptionally mediocre. Gain will be marginal. Output impedance will be higher than optimum for most speakers, but not so terrible as to end all life as we know it. A split-load phase inverter tends to have low distortion, but is utterly dependent on the stage driving it. In this case, said stage uses a tube of stunningly mediocre linearity and marginal mu. On the plus side, the classic grounded-cathode-dc'd-to-a-split-load is classic for a good reason- it's an excellent topology.

Putting religion aside, when all is said and done, a design like this will generally benefit from feedback, which means you want to put a lot more gain up front and take some elementary precautions to minimize overload recovery problems.

And don't be afraid to run more current through that split load- the low current you have there may not charge the grid capacitance of the output tubes as quickly as you'd like.

For reasonably sized grid stoppers (below 10k or so), Miller capacitance won't be a huge issue. Values are very noncritical. 99.99% of the time, 4k7 will be fine. So will 3k3 and 6k8. Got it?

And finally, I'll repeat the advice I give to anyone who wants to get serious enough to design their own tube amp- buy "Valve Amplifiers" by Morgan Jones, easily the best currently-available book on the subject. Read it thoroughly. He works through every one of the issues you've brought up and in great detail. It will cost you less than a set of output tubes.

 

[jayme]

'How to configure the quad-quasi-quantum-octode-megaquench-converter for construction of an RF cookie baking oven (volume XXXVII)'

I tried to search for "RF cookie baking oven", but didn't find any results. Now there is a tube-based project I can finally get the wife to buy into!!!

First, as SY says, buy Morgan Jones "Valve Amplifiers" off of Amazon. I've read it cover-to-cover about 15 times now, and learn something new every time as I finally tie together a problem I am dealing with and the advice in this book. It truly is a seminal work. And you won't even realize how SPOT ON it is until you've done a lot of debugging.... Plus, it covers volume pot impedance, cap bypass, etc...all of your questions below.

Second, a CRC or CLC supply is a LOT more complicated than it may seem...there are resonance issues, impedance issues, etc that are not apparent at first model with PSUD. My CLCLC filter (ASC caps and Lundahl chokes and power transformer...about $450 worth of parts) modeled great in PSUD but has proved to have motorboating characteristics in the real-world application. (SY has suggested adding a regulator stage...I am exploring that option now). I've come to realize that a well-designed PSU is a lot harder than just throwing chokes and caps in a chain.

Third, in terms of resistor wattages...calculate what the worst-case current through the resistor could be if a "failure" happens, and multiply by the worst-case voltage if a failure happens. For PSU resistors, model what happens if the load suddenly drops to 0mA or rises to 2x above your nominal current draw....PSUD will calculate the voltage drop across the resistor and the current though the resistor at these points. That, at least, is how I've approached sizing resistor wattages required.

 

[d2134]

The PSU dose not work; you have negative voltage across the CCS.
You have 412Vdc at filter input (294*1.4), you have 594Vdc (.421*1400) voltage drop across the resistors.This result -182Vdc at output. This are handcalculations.

In a SPICE-like program the values above are: 399Vdc, 589Vdc, -190Vdc.

On the other hand, the power disspation on the resistors in the CRCRC filter is huge. You have to use much smaller resistors.

 

[PeteN]

Great stuff, thanks Will look into that book asap, if it's as good and cheap as you say, sounds like a steal!

I guess it was kind of predictable that a "throw a crc or two together and model it" PSU wouldn't fare too well! \

I soppose it's daft to pose further questions without reading the book, so I'll hold off for now!

(NFB was something I meant to ask about, actually, but it can wait for now)

I'll get the book and (now that the exam period is drawing to a close!) have a tinker with some numbers, and see what I can come up with (actually it'll probably be coming down and with a headache!)