• 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.

ECL86 preamp critique please

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
ecl86 preamp critique please

ok - help required at a pretty basic level.

I fell in love with poinz's mobius preamp - its elegant simple design (visually) and its easy to follow (sorta) electrical design. Way cool.

I had a similar presenting problem - a largeish MOSFET output power amplifier (NZ build McLaren 701) is lurking around my place and is underutilized cos it has no preamp. I'd built a set of three 12ax7/el84 SE amps for my kids recently and caught "the bug". So a tube pre is the only reasonable option...

Heres the rub - I tend to play out of my depth, so immediately I was drawn to all sorts of complex or esoteric solutions - no simple 12xx7 grounded cathode fix me me - no sir! Immediately it was clear - I had no idea what to do or how to do it!

Then recently I was led to valvewizard who, after the complexity of John Broskie's fantastic but waaaaay deep stuff or the once over lightly at boozhound (no disrepec' here - that site certainly helped understand the kids el84 amps), well the wiz is able to put over the right combo of maths, intuition and real ife application. For me.

Anyway, the wiz has a real simple explanation of how to apply feedback on a triode stage. Up until then it seemed a black science driven by incantations and chicken entrails. Not true - its simple maths - and the idea is just application of parallel circuits and ratios. Whoa.

And then the wiz describes cathode followers!!!! What a guy! Same approach. I have a new hero.

*Sits back, fans self with hat and recovers*

OK, so now I'm real excited. I have the parts (I think) and the knowledge (kinda) to produce the solution - a preamp using 2 ecl86 tubes, the triode sections providing amplification of the line signal with local feedback to tame any nasty stuff and to temper the triode's over-exuberant mu. THe output goes to the pentode sections configured as a cathode follower to give plenty of current drive for the butty mosfet power amp.

Known issues: Gotta lift the heaters to around +80v to get within the required heater-cathode rating.

Unknowns: How do I separate the HT DC from the feedback circuit? I guess a cap of some sort, but how to work it out?

I appreciate there are no values shown - I figure I may be scrapping the lot on your collective advice, so why bother with all that calculus?

Feel free to criticise liberally but not along the lines of "Oh but a ecl86 is sooooo passe, everyone is using the NOS Unobtainium Deluxe triodes hand-built by Ukranian virgins and available only by invitation..." Just the electrical design stuff please.
 

Attachments

  • cathode follower preamp.pdf
    10.9 KB · Views: 739
aard, problem 1 is that your feedback level (and hence distortion spectrum) varies with volume control setting. This topology choice also forces you to run things through two more capacitors than necessary.

Although I've never been able to measure the effect of LED nonlinearity in an actual circuit (the dynamic impedance is so low that any changes with current are down in the noise), the 1mA or so that your first stage will run does put it on the high slope portion of the Vf/If curves. You can ameliorate that by running a resistor from B+ to the cathode, sized to run an extra 8 or 9 mA through the LED.

If you're ambitious, you can increase the feedback and remove any question of LED nonlinearity by using a CCS as the first stage plate load. That will also give you better power supply rejection.
 
I think overall its a terrible approach to the preamp problem.
The ECL86 is an excellent valve in the right place, but the front end valve has much to much gain which forces you to use a lot of feedback to bring it down. Final result will be a dull sounding circuit little better than a simple opamp.
There are an infinity of better designs out there,

Shoog
 
'Vark,

I'm in agreement with Shoog. The triode in the 6GW8/ECL86 is "identical" to those found in the 12AX7/ECC83. The mu of 100 is not well suited to line stage service.

Let's back up and start with the SS power amp that is to be driven. What's the drive voltage needed to achieve full power O/P? What is the I/P impedance of the SS power amp? Those numbers tell us what sort of circuitry is appropriate, in this particular situation.
 
Originally #3 posted by SY


Although I've never been able to measure the effect of LED nonlinearity in an actual circuit (the dynamic impedance is so low that any changes with current are down in the noise), the 1mA or so that your first stage will run does put it on the high slope portion of the Vf/If curves. You can ameliorate that by running a resistor from B+ to the cathode, sized to run an extra 8 or 9 mA through the LED.


Hello SY,
have a look at the datasheets of LED's. ;)
You find forward current vs. voltage curves there.
From the curve you can calculate the dynamic impedance.
I always put a bypass cap in parallel to LEDs in
stabilizing circuits to get no "LED noise".

Kind regards,
Darius

edit: You are right, I tested it out, noise is not a problem.
thanks. :)
 
FWIW,

why don't you just build the moebius? Its a very good sounding preamp with few parts, an easy to achieve PS and the tubes are cheap. Signal goes through 1 resistor and 1 cap IIRC. Very simple design with few parts, much simpler than an aikido too (of which I have 2 and think are really great).

Another option would be a 24V aikido for many of the above reasons.

What more do you want??

Fran
 
Re: ecl86 preamp critique please

aardvarkash10 said:
Unknowns: How do I separate the HT DC from the feedback circuit? I guess a cap of some sort, but how to work it out?

R9 x C4 > R8 x C1

Darius
 

Attachments

  • ecl86_lineamp.jpg
    ecl86_lineamp.jpg
    36.2 KB · Views: 751
whoa! Babelfish!!!!

Is there any better tool for gut wrenching belly laughs??? Or, as it may yet be translateed into some obscure language, "colon-spannered digestive breathpulses"? Thanks for that lead my antipodean adviser. However, poinz's solution is so elgant in both design and execution, its hard to beat. As I earlier noted, I have a quartet of 6CH6 tubes that are just awaiting triode curves so I can impliment them in his design (tight-as kiwi here - use what ya got...) in place of the 6AQ5 he designed for. This has two advantages: I get to learn more, and I use stuff that cost $15 NZ.
 
Re: whoa! Babelfish!!!!

aardvarkash10 said:
Is there any better tool for gut wrenching belly laughs??? Or, as it may yet be translateed into some obscure language, "colon-spannered digestive breathpulses"? Thanks for that lead my antipodean adviser. However, poinz's solution is so elgant in both design and execution, its hard to beat. As I earlier noted, I have a quartet of 6CH6 tubes that are just awaiting triode curves so I can impliment them in his design (tight-as kiwi here - use what ya got...) in place of the 6AQ5 he designed for. This has two advantages: I get to learn more, and I use stuff that cost $15 NZ.

Hi,
are you talking about this one?

Darius
 
Referring to the original circuit:

You could reduce the number of coupling caps to 1 with a few modifications. For starters, you could try to chose an operating point for the triode, which alowes you to directly couple it's plate to the grid of the triode connected pentode used as a follower. This has the consequence of reducing available Vkp, and possibly increasing the dissipation in the catkode resitor of the follower, but it might well be worth it.

Secondly, you can dispense with the input cap. You could put one in front of the pot 'just in case', which could also be bypassed by a switch or jumper. At this sort of bias condition, there should really be no problem with grid current.

Thirdly, your feedback could then go directly from the output after the single remaining coupling cap. The problem with separate coupling caps for feedback and output is that the feedback becomes lower under a certain minimum frequency determined by the feedback network and cap, the added catch being that the feedback network is somewhat variable with the setting of the input pot. On the other hand, the output cut off frequency is determined by the chosen output coupling cap and the load, which is possibly unknown. A problem arises if the lower cut-off frequency of the output ends up being lower than that of the feedback cut-off, in form of a gain rise at low (possibly infrasonic) frequencies. By using the output coupling cap as the feedback cap, the two lower cut-off frequencies become one, so there is no such problem. A slight hitch is that initially (uppon coupling cap charging on start-up) the feedback network provides a path for the cap to charge, so this circuit could do well with a muting relay (that shorts the output to ground initially).

Finally, for the price of an extra resistor and cap, and (unfortunately) some increase in output impedance, you could implemnt a 'bootstrap CCS' for the input triode, by splitting the plate resistor in two and coupling a bootstrap cap from the connection of the two-part plate resistor to the cathode of the follower. This will in fact increase gain and linearity but also increase feedback, which will already have to be quite high, so take this idea with a grain of salt or two...

PS - the gain of the triode section approaches levels where you can assume that the input impedance of the whole line stage is close to the value of the input resistor. This means that at full volume, the input impedance is said resistor in parallel with the pot. In order to keep Mr. Miller at bay, the input resistor cannot be very large. Still, because of this paralleling, the input impedance varies considerably with pot wiper position, for typical values it's a 1:2 ratio. You can actually make this better and improve pot tracking by connecting a resistor equal to the input resistor between the pot wiper and the top end of the resistance element (the input of the whole amp). This lowers the input impedance to input res in parallel with pot res on both ends of the pot (but you have to count on this as the minimum input impedance anyway), but the variation of the impedance with pot wiper position is much reduced, a ratio of 3:4.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.