Replace 1N5408 Diodes with Hexfreds in a Tube Amplifier

[Smoky]

Hey guys, I'm in the middle of gutting two huge mono tube amplifiers from Cary Audio.  I'm repainting them, and to do that, the chassis and transformer covers needed to be stripped/removed.

The amplifier has two large potted rectifiers for the B+ (~1035 VDC, part# HV8-1) that I'm not touching but there are five 1N5408's for two other power supply rails (+285 VDC and -200 VDC) that are mounted awkwardly across two separate terminal tie strips.

My plan initially was to mount new 1N5408's on a piece of .060" epoxy/fiberglass sheet with tin-plated brass eyelets. This would clean-up the rat's nest of wires within these amplifiers. I'm also doing this for bundled capacitors in other areas. I see wasted/excessive lengths of wire all over the place.

I heard about Hexfreds, and while I'm at it, should I use them?

Here is one from Digikey:

https://www.digikey.com/product-detail/en/vishay-semiconductor-diodes-division/VS-HFA06PB120-N3/VS-HFA06PB120-N3GI-ND/4810472

[wraper]

Waste of money and inferior forward voltage drop. They are not meant for rectification at low frequency, you can use them but there is no advantage.

[Smoky]

Thank you Wraper, that's all I needed to know.

Here's a shot of the schematic of the two supplies (C-200v and third B+285v).

[TimFox]

Why is a high current tube (6CA7) used in the “tail” of a low current 6SL7?  Single-ended Class-A operation of the 6CA7/EL34 is typically at 70 to 100 mA, while a 6SL7 usually runs 2 mA or less per triode.  At first, I thought it was a typo for 6AC7.  Both tubes have separate G3 and cathode pins.  The cathode current of the pentode (based on the voltages at the cathode resistor) corresponds to that level of current at the 6SL7.

[floobydust]

You don't really need 3A diodes there, it maybe from a common mistake upsizing from 1N4007 to 1N5408. It doesn't work at high voltages, you get worse reliability. 1N5408 sucks, they are very slow tt ~5usec. The fast dV/dt near zero-cross and reverse recovery can generate a lot of EMI from their switching. Just put a portable AM radio near the chassis wiring and see for yourself.
I've been using faster "soft recovery" UF5408 75nsec and have not tried ultrafast SF5408 50nsec. Tempted to try SiC but I find an RC snubber across the power transformer secondary works very well to the mitigate rectifier EMI at low cost, if you are improving the design.

[Smoky]

I restored the amplifiers "electrically" a year or so ago.

The paint job left something to be desired. A black metallic with every flake of the rainbow sprayed over textured powder-paint on the aluminum transformer covers and the grounded coarse scratches of the steel chassis. No thank you!

I have one chassis already chemically stripped and sprayed with two coats of epoxy primer. The dents in the aluminum transformer covers have been repaired and hit with epoxy too. All pieces are ready for two to three coats of black urethane surfacer.

The project will take its time. But it's fun 

[Smoky]

Good to hear from you FloobyDust!

When I dug into these things the main goal was to reach the lowest amount of noise at idle.

When it was all said and done, both amplifiers averaged ~520 microvolts of noise floor. Close to silence out of the speakers.

I just thought if there were any tweaks to be done, now is the time.

[floobydust]

Some of these boutique amplifiers aren't really that great.
Here is Mr. Carlson's approach worth watching, although I would never hold a live choke in my hand...

I find cleaning up RFI in the power supply has sonic benefits in any amplifier.
While it doesn't make it to the loudspeaker outputs, depending on the chassis it can radiate or otherwise be swimming around the pre/driver sections. It would not be present with old rectifiers say 5U4's and so old designs had nothing to deal with it, but modern stuff with slow HV diodes seem to need help.

On the open chassis, holding a scope probe inches away can show you some hotspots.

[Smoky]

Here's one of the cool discoveries when I put things back together for the first time. It had to do with the installation of the new front panel milliamperes meters.

In one of the original meters the pointer was stuck, so I searched and searched and found two NOS edgewise bias meters that would work.

The problem I thought was, how would I zero the pointer once they were installed? They are trapped in the chassis with no access.

Get this, Cary Audio aligned the center of the octal socket of the 6SL7 perfectly over the front panel bias meter adjustment pot!

[Smoky]

*Now this response only relates to the Six Pac Amplifiers FloobyDust is referencing. Not the CAD211FE's related to this thread.*

FloobyDust, the Six Pac amplifiers are bad to the bone. I own a pair.

Here is a shot of the bottom of the Six Pac amplifiers after I rebuilt them. This was just after the repair of my first TDS420 oscilloscope

I lifted the choke onto a bracket fashioned out of aluminum angle and pivoted the choke ~30 degrees.

When they were done, both amplifier's noise floors averaged ~620 microvolts. With the feedback switched to -10dB, the amplifier's noise floors dropped to ~440 microvolts.

I think this bested Mr. Carlson's attempt by ~400 microvolts.

They are my pride and joy of the electronics projects completed so far.

[Smoky]

Do you remember in Mr. Carlson's video when he used his Tektronix 547 oscilloscope to capture the "ringing" in the Six Pac amplifiers?

Well, I saved a picture of how well my TDS420 with a P6009 probe captured it too. Good stuff

And thank you FloobyDust, I'll order some of the UF5408's!

[David Hess]

You don't really need 3A diodes there, it maybe from a common mistake upsizing from 1N4007 to 1N5408. It doesn't work at high voltages, you get worse reliability. 1N5408 sucks, they are very slow tt ~5usec. The fast dV/dt near zero-cross and reverse recovery can generate a lot of EMI from their switching. Just put a portable AM radio near the chassis wiring and see for yourself.

I've been using faster "soft recovery" UF5408 75nsec and have not tried ultrafast SF5408 50nsec. Tempted to try SiC but I find an RC snubber across the power transformer secondary works very well to the mitigate rectifier EMI at low cost, if you are improving the design.

I have had the same experience with standard recovery rectifiers in power supplies for audio equipment and some test instruments.  Some standard recovery rectifiers, and it seems to vary by manufacturer and even production lot, act like a snap-off diode producing RFI with a repetition rate of the line frequency.  High voltage parts should be worse owing to their PIN construction but I have encountered it with low voltage parts as well.

If I do not want change the diodes, then adding a snubber capacitor across the diode with several times the diode's capacitance has always worked to suppressed the noise.  Another recommendation I have seen is to add ferrite beads.  Fast and soft recovery rectifiers never seem to have this problem.

On the other hand, the problem is not universal so try the 1N5408s and if there is a problem, you should know pretty quickly.

... but there are five 1N5408's for two other power supply rails (+285 VDC and -200 VDC) that are mounted awkwardly across two separate terminal tie strips.

My plan initially was to mount new 1N5408's on a piece of .060" epoxy/fiberglass sheet with tin-plated brass eyelets. This would clean-up the rat's nest of wires within these amplifiers. I'm also doing this for bundled capacitors in other areas. I see wasted/excessive lengths of wire all over the place.

I like the old style terminal strips shown here and point to point wiring:

https://en.wikipedia.org/wiki/Terminal_(electronics)

I heard about Hexfreds, and while I'm at it, should I use them?

Hexfreds qualify as fast and soft recovery but the TO-220 style packages would be inconvenient for point to point wiring.  I would probably use axial leaded BYT78s in the SOD-64 package.

[Smoky]

Thank you David.

Here is a photograph of four of the original 1N5408 diodes straddling two terminal strips in the CAD211FE amplifiers.

I had the intention of mounting the four new diodes and the 47uf 450v capacitor, that you see laying on its side, on one small board.

My only intention was to tidy things up since the amplifier performs very well.

I do have plenty of NOS Sprague strips for the new UF5408's or BYT78's. I'll buy both types.

[Smoky]

Just one more comment on the AES Six Pac amplifiers.

Mr. Carlson used a slightly lower value of capacitor to knock down the ringing in his amplifiers. I think it was .47uf?

Well, the two amplifiers I worked on required a slightly higher value, and like Paul, I experimented with several.

I also used a  Sencore SC61 along with the TDS420 to monitor the results. I'm glad I took a lot of pictures! It is really cool to go back and look at this. I think I understand more of it now than when I worked on it

The first picture is of the "ringing" prior to adding any capacitance to the choke.

The second picture is the result after installing an Epcose/TDK .68uf 1kV film capacitor. The VPP rose very little if any.

The capacitor was mounted using automotive attachment tape to the side of the chassis.

Here is the capacitor from Digi-Key, but I'm sure your amplifier may react differently:

https://www.digikey.com/product-detail/en/tdk-electronics-inc/B32654A0684J000/495-75758-ND/1648098

[trobbins]

Whoa, the 285V rail only supplies a few mA.  The 329V rail supplies about 100mA.  You haven't shown the power supply schematic.

For those levels of dc load current, a UF4007 is suitably over-rated.  Rectifier noise etc gets worse with higher current rated diodes - so the UF5408 will be worse in that regard.  And there is no need for any snubbers across that diode, as snubbers will only make rectification noise worse when using appropriate diodes and wiring layout.  Make sure your wiring layout of the power transformer secondaries, rectifier and filter caps is 'best practise', otherwise it doesn't matter what diodes you use - perhaps worth posting the power supply schematic and some photos showing layout.

Just my 2c.

[Smoky]

I do see a transistor on the third B+ rail, so I imagine it's regulated.

After reading a few datasheets on the recommended diodes, they all seem to have slower switching times once you cross above the 500v threshold. It makes sense to select a diode that does the job adequately.

The 1N5408 diodes in this amplifier were factory installed.

[floobydust]

Odd to see oversized rectifiers but with retail price of $22,000 just put them in for insurance.
Realistically, transformer HV windings and choke will be 25-65R each so inrush is already moderated for the huge 1,500uF 450V filter capacitors.

I'm a bit suspicious of that 2R 5W resistor for the 1,000V rail. It does nothing for inrush. Even for monitoring current on the 1,000V rail, you would move the part after the filter caps. It seems to be at most a tiny 4uH inductance there.
What I do is take a few capacitance measurements of the power transformer. Primary-secondaries, then the chassis etc. To see how it is wound and isolated the windings truly are. Some are cursed, so you get large capacitive coupling from one winding to the other, enough to induce unexpected transients and drama.
I use RC snubbers across each secondary (not across each diode) based on scope measurements, to tame leakage inductance from killing the diodes or making RFI. Ferrite beads downstream did not work as expected.

[Smoky]

Each one of those CAD211FE mono amps weigh 90 pounds a piece too!

And of course, when I seen that Chinese ceramic sockets were used in both the CAD211FE's and the Six Pac amplifiers, out they went.

I fitted the CAD211FE's with NOS E.F. Johnson tube sockets dated from the 1950's. The design of the Johnson 211 socket contacts are far superior to anything on the market today. Glazed Soviet octal sockets with silver-plated contacts were used for the 6SL7 and 6CA7. They're dated around 1962.

The Six Pac's got un-glazed Soviet 8 and 9 pin sockets with silver-plated contacts from around 1966.

And since the chassis are deep in the Six Pac amplifiers, I took the measurements of the center-line spacing of the sockets and soldered-in 90% of the wire connections while they were screwed down to the bench.

Lots of work but it was well worth it 

[Smoky]

FloobyDust, here's the 2R 5Watt you spoke about and it's a prime example of the way things are oddly wired on the tie strips.

All three of those tabs you see on the left are grounds. Two wires are just jumpers. Plus, they have another black wire that ties all of the grounds in the chassis together.

There is another area in the chassis where I can delete five wires if a place the few components involved onto a small PC board and have all connections converge at one point.

So what I'm going to do, and since I have these little press-in silver-plated posts and epoxy sheets, is wait for the tin-plated eyelets to come in and fabricate some groovy shapes to fit some of these components on. I think it will be cool. it will have the sense of point-to-point with radial capacitors looking like they're mounted correctly with their leads below and leaded components next to them mounted between posts on top.

Btw, all of the wire that they used in this amplifier is 19g Kimber TCS.

There is also a 220nF 3% 630v film capacitor mounted on top of each 1,500uf 450v capacitor that's not shown in the schematic.

[trobbins]

If the amp is oldish then UF5408 may not have been available.  In general, a well made amp doesn't 'need' fast rectifiers - the use of fast rectifiers is more a recent attempt to 'upgrade' performance but where nearly every upgrader doesn't actually test to see if any performance attribute has change - the upgrade is more about kudos or bragging rights for the upgrader.

That said, technology and awareness of improvements do mean that if you have the incentive then using a higher speed diode, and as floobydust indicates, measuring the glitch performance of power transformer secondary windings and applying a tuned snubber (aka using the quasimodo type test procedure) can minimise the influence of secondary winding leakage inductance on the nearby environment.

The 2R resistor may have been inserted for a few reasons, such as a form of over-current protection, a current monitoring point, and a peak in-rush current limiter.  How much influence it has on in-rush would depend on measuring DCR of windings and doing a PSUD2 assessment.

Your scope is having a hard time measuring the hum/noise on the output - given the frequency that is resolved.  A soundcard/spectrum software type measurement along with a meter would provide a much more confident measurement scheme, especially if you were aiming for before/after type measurements.

I would caution any change in 'ground' wiring - that is where a good awareness of current loops and hum/noise management can be a big issue - there may well be a good reason for using many terminals to manage grounding.

The use of low capacitance bypass caps across large electrolytics is not recommended imho for modern electrolytics.  A good new electrolytic will have a low and sustained minimum impedance that is dictated by its ESR and ESL over a very wide frequency range, that typically extends out beyond 100kHz.  A small bypass cap may provide a lower impedance high-Q notch somewhere in the MHz range - but that can be a detriment, as it introduces a risk of tuned noise currents.  The aim was I suggest to provide a better signal path for higher frequency output stage large current signals - but the very large capacitor bank with its large loop area and location is not really where you want to loop such higher frequency currents.

[Smoky]

Thank you for the comments Trobbins.

We will do those assessment tests on the CAD211FE transformers for sure, I'll just need help on how to do it

I can assure you too that there will be no change in the ground wiring. I'm simply speaking of components that are located in such a way that their size or location required special wire routing or lengths. Within the CAD211FE and Six Pac amplifiers, the ground wire routing seems to follow star-type grounding rules.

The small 220nF 630v ClartyCaps bypass capacitors on top of those 1,500uf 450v were installed at the factory. The label on them reads "custom made for Cary Audio."

Now for a short story on the TDS420 measuring the noise at the speaker terminals of the amplifiers. An 8-Ohm speaker was installed at all times. Looking back at the pictures, I was even using a 10x P6138A probe. It's all I had.

It was two years ago that I brought back the TDS420 from the dead, my first ever electronics project. My next project was the repair of the Six Pac amplifiers. The first time I plugged the amplifiers in, noise was coming out of the speaker louder than the pair Paul Carlson worked on. No wonder the previous owner stuffed them into the back of his closet!

I started by gutting each amplifier one at a time. New capacitors, resistors, and the rearrangement of the choke came next.

The time came to power-up the rebuilt Six Pac's. I used a variable isolation transformer and the speaker terminals were monitored by the TDS420 and a Bose speaker. I used an external milliamp meter to set the bias at ~220mA.

What came next was horrifying! The TDS420 and my ears were hit with over 29mV of noise coming out of the speaker as I scrolled up the voltage of the isolation transformer. My heart dropped into my stomach! I shut it down. I didn't know what to think. But then something I recalled reading popped into my head. It was an article discussing capacitors and how they take time to burn-in. So I went back to the bench and slowly brought the voltage back up while tolerating the noise out of the speaker.

It was amazing. The oscilloscope read 29mV, then 26mV, 23, 21, 19,..., 15, 12, 8 and after a few minutes it was hovering around 3, then down to 2mV. When it dropped below 1mV, I nearly fell to the floor

And about the frequency, from my experience, the TDS420's and 460's jump around a bit. The frequency doesn't always lock firmly. Maybe it's due to the acquire-mode setting that is used 

Since then, I came into possession of an HP 35660A DSA. It has its own signal source too. Maybe we can use it to test these CAD211FE's once they're done?

Here's a shot of the new Kemet filter caps I used in the Six Pac's compared to the ones used in the CAD211FE's.

[trobbins]

That HP35660A looks modern enough to be similar to a soundcard's dynamic range and noise floor.  My older HP3560A is nowadays quite limited in its dynamic range and is ugly to interface to, so my goto measurement system is a USB soundcard and REW.

Using a 1x probe would certainly raise the signal level on your TDS420, and allow it to make a better basic measurement of rms level as well as give some FFT based frequency spectrum indication - well worth getting to know what you've got.

I'm not sure the hum you measured was just due to new caps.  Certainly if the caps had not been used for a few years then they can draw additional current whilst they reform back to normal, but I think the elephant in the room are the output stage valves, which can take time to thermally stabilise and any PP differences in time may show up as hum as the amp doesn't include feedback that would typically attenuate that hum.

[floobydust]

I've stopped using those old Cinch Jones terminal strips, although they are a time trek with their bakelite aroma, they just don't hold many parts and can arc over past a few kV. I've switched over to glass epoxy Keystone Turret Terminal boards for old point-point stuff. The corner posts are not always usable due to the mounting bolts.

I would check the important (higher current) grounding. There's at least 3 ways to go about it, and sometimes it's a mess. The 2R resistor strip looks like somebody had issues and was playing around. He may have tried to intentionally add hum to a stage to cancel it. The three ground lugs seem to be the star point but the 845 cathodes (after 100mA jack) should return as a wire right to the PSU filter cap (-) not the chassis. It's a lost art that Leo Fender had perfect in the 1950's.
Maybe see how it Cary did it.