Please critique my plans for a WE91 clone.

sgerus said:

For my next amp project, I’m planning on a circuit similar to the WE91.
OK. It starts with a pentode, and that where the similarity ends.

The input/driver is a 6sg7 with a 0c3 to regulate the g2 voltage. The power tube is a JJ2A3-40 (300B with 2.5V filament)

Here are some specific questions, see the attached schematic.

1. Is the circuit for the 0c3 correct?

2. When using a 0c3 to set the g2 voltage of the pentode, do you need a capacitor from g2 to the cathode?

3. In general, is the 0c3 worth the cost (heat) of the 15ma it burns to regulate the g2 1-2 mA

4.For the pentode, I want an octal based tube, preferably glass. I’ve looked at:
6j7gt at 3mA (cathode)
6sj7 at 6.25mA
6sg7 at 8.1mA
On paper, the 6sg7 looks like the best choice, (lowest distortion, very high voltage swing, and the most current)
Any thoughts on my choice of pentodes?

5. Is the power supply overkill…. IE the separate “g2” voltage for the oc3

Thank You -Scott

Click to expand...

I would probably choose the 6SG7 and then the 6SJ7 in that order.

Using an OC3 as a screen regulator is a great idea, shunting it with anything more than 0.01uF - 0.022uF probably isn't, at 0.1uF I would be concerned that the gas tube and cap will turn into a relaxation oscillator.

I would use a standard 300B rather than the JJ 2.5V version (2A3-40) as this will give you a lot more flexibility in terms of future tube choices. (And sadly JJ seems to be having some pretty significant QC issues with larger power tubes at the moment.) Using a 5V tube will require dc filaments (ccs recommended) for low noise - the one real advantage of the 2.5V JJ tube.

I haven't looked closely at anything else, but it looks like a reasonable starting point.

Edit:

I spoke too soon - I found some other things. (refinements mostly) The cathode bypass cap and the ultrapath cap ought to be higher in value for good bass performance. I'd recommend 80 - 100uF and for effective ripple cancelling the ultrapath cap should be sized using this relationship
C_ultrapath=C_bypass/Mu +1 in this case with 100uF a 20 - 22uF gives you a significant amount of cancellation of supply ripple in the output stage. In a similar amp I did a couple of years ago Spice predicted something like 20dB of additional ripple suppression and I got something > 15dB - a rather worthwhile improvement.

The quality of the ultrapath capacitor is quite important, I used a 20uF Solen with a 100uF ASC motor run as cathode bypass cap. A better ultrapath cap than the Solen is probably a good investment.

Gas tubes can be quite noisy, hopefully this is not an issue here, if it is a zener may well be quieter, and can be bypassed with a honking big cap if desired. Sometimes RC decoupling can be used, but R has to be small to minimize IR drop with varying screen current which means that to prevent relaxation oscillations the cap still has to be pretty small. It does establish a predictable HF pole which is a good thing.

You can't use a cap of meaningful size on the screen to the cathode or ground with a gas tube because of the oscillation issue, and in any case the source impedance of the gas tube is extremely low. You do need to make sure the cathode is well bypassed to ground for best linearity.

In terms of OC3 current you could probably get by with a little less than 15mA, but this or even higher current will lower the internal impedance of the gas tube. You have plenty of voltage margin so the tube should "strike" reliably.

I don't see any particular problems with the psu design, you might want to check it in psud if you haven't done so already..

Arcol makes great little heatsinkable aluminum cased power resistors for chassis mounting that cost a lot less than the Dales they compete with. They're available from Mouser.

Good ASC motor run caps for cathode bypass and power supply stuff is available from midori meadows on eBay. (Highly recommended.)

sgerus said:

Thanks Kevin,

I committed to the 2.5V filament for the power tube (custom made T1 on order).
So that leaves me with the Sovtek 2a3 runing at about 21 watts, take my chances with the JJ (that was original plan), or spend about $375 a pair for EML 2a3-s or KR 2a3.

I will change the 0C3 to run at 10mA with a .01uf cap.

What brand of capacitor would you recommend for the “ultra-path”… in place of the Solen?

Also, what do you think about just using the 50-100uf ultra-path capacitor and omitting the cap on the cathode bypass resistor?

Click to expand...

I tried just the ultrapath configuration and found I preferred the two cap configuration, and with a single cap you loose the ripple canceling inherently possible with this configuration.

In terms of something better than Solen I was thinking the Clarity Cap SA would be one possibility, the Erse Pulse X another - both are available from a variety of sources like www.diycable.com, and Clarity from www.e-speakers.com The nice thing about these caps is they sound good and aren't very expensive unlike many other boutique brands out there. ASC motor run caps on eBay are cost effective and pretty good sounding in this configuration - a bit soft on the highs if used by themselves.

Looks like you have some good options for later with both EML and KR making suitable replacements for the 2A3-40..

Sounds like a cool project, be sure to post details as it all unfolds, pix of course would be great.

Kevin

Hi I wouldn't go with 6SG7 as these are "semi remote" pentodes

How about an EF37A. High quality, lo-noise drop in octal replacement for a 6J7 with grid cap also. My Mullards have nice red paint on glass envelope.

Quite different is the 6AG7 metal octal or the 6CL6, 9 pin glass miniature. Those swing large and are used to drive video transmission lines at 75 ohms. I was planning to try the 6AG7 as a driver in a PSET multiple 6080 design.

Hi Scott,
The proof of course is in the pudding.. Took a quick look at your schematic and your design looks good to me.

Since you are limited in plate dissipation I think your chosen operating point for the EF37A is a reasonable one. (Based on your schematic values, not the ones listed in a previous post.)

Power level and distortion are quite good, particularly when you consider this is a two stage design and the 2A3-40 is really closer to a 300B in drive requirements when operated at these voltage.

How about some pix of your new amplifier?

Nice looking amp Scott, what is the frequency response like?


Thanks

Andrew

sgerus,

Your amp looks terrific! Those red EF37`s are a nice cosmetic touch.

Sgerus,

Nice job on the amp... having used gas regulators for many years, they can be somewhat noisy. I would probably be inclined to decouple it from the screen with a resistor and use a cap to the cathode. Also, as the "other" Kevin pointed out, be careful on the cap in parallel with it, I've always used 0.01uF and never had a problem. For very light current use, the 5651 is an interesting tube... but it has a narrow (current) operating range and has typically been used as a reference in regulated supplies that require tighter regulation than most higher voltage/current VR tubes used as shunt regulators.

Regarding the hum level... 1mv is pretty high, especially for a few watts output. I prefer AC filaments when possible and have implemented a split balance technique for all of my 45 and 2A3 designs. I prefer to use a center-tapped filament winding for the DC path to ground, ie, the bias resistor. This provides a very low impedance thru the winding to ground. For AC balance, I use a padded 100 ohm pot (use a pair of 12 ohm resistors) which yields a 20 ohm balance pot without the knife-edge adjustment. From the wiper, just use the bypass cap by itself to ground. This splits the return path, DC is fixed at center, AC is adjustable.

Using this technique, I've measured output noise with good 45 tubes below 150 microvolts.. even fair ones are generally below 400 microvolts. For 2A3 tubes, I've measured below 300 microvolts on good ones, but I have some NOS RCA blackplate versions that are around the 150 mark. I also posted a DIY 45 design on the MQ forum....

Measured specifications taken at 1-watt output:

20Hz - 40KHz within 1dB (James H3115... expect the RH-40 to do better)
THD less than 0.4% (less than 1% at 2-watts output)
Signal-to-Noise ratio: 80dB or better (referenced 1-watt/8-ohm)
Output noise ~ 250 micro-volts
Gain - approx. 11dB (775mv for 1-watt output - 2.83VRMS into 8-ohm)

Note: the above design uses the split balancing technique above. I hate cross-posting, but here's a link to the PDF schematic:

http://www.q-audio.com/images/lc45_rh.pdf

Regards, KM

HI Sgerus,

Is your driver a 5842 or 5842WA? iirc the WA version will bias up higher, around 3V I think versus ~1.5 for the 5842 and I think you can actually get away with driving a 300B with the WA version.

I was using them (5842) on a SE 2A3 IT amp years ago when the 417 was all the fashion but the 5842 was cheaper.
I still think it is a good driver once tamed.
The 6c45 is also a candidate but I think is even more prone to oscillation.

I would have thought you would have more bandwidth on the top with the 5842 although it would be just semantics in my case as I can just hear around 15Kcycles or could anyway last time I checked many years ago.

What opt's are you using?

The JJ 2a3-40 are beautiful to look at. Did you compare them soundwise to any others?

Are the EML's and KR's still manufactured or NOS? I did not know they were still around as I have not heard of either of them for years.

Thanks for the update on your project.

Andrew