Mullard 5-20 high DC Voltages

[Diabolical Artificer]

It's a sod having to scrap work, especially after you'd laid out that circuit board so nicely, just one possible fly in the proverbial oitment: those tag boards can cause issues if your not careful if the final chassis is going to be metal - lot's of potential gnd loops.

As an aid to building your 5-20 it might be beneficial to look at the 5-20 like clone circuits for ideas/tweaks.

Andy.

[Kei-1986]

Yes, it was quite annoying scrapping the first design, but I did consider it a prototype to prove the concept. It struck me as more sensible to just chalk it up to experience and rebuild to spec and then debug that try and debug something that could potentially require more significant modification possibly at the expense of sound quality This Mk2 version will be just as neat if not neater. Having never undertaken a project like this previously, it is a quite learning curve. I've usually stuck with repairing old amplifiers rather than scratch builds. The only things I'd scratch built prior to this were speakers.

One potential ground loop I have spotted is the IEC input socket as it's metal clad. The standard mullard design calls for the chassis to ground at the input socket rather than the mains input. I don't see the turret board causing any issues as it is isolated from the chassis. The ground bus will also be insulated from the chassis. I'm keeping my fingers crossed that I'll get the grounding right. One of the things that crossed my mind as to the possible cause of instability was the lack of a grounded metal chassis.

The main clone I looked at originally was the beam echo dl7-35. I also looked at the Leak Tl25 but that seems to differ in quite a few ways. The Claus Byrith "4-30" modified version was the main design I looked to for inspiration for modifications/tweaks to the design. The only change I could see as being really beneficial was to change EF86 from pentode to triode to reduce the gain.

[rontech]

I always use IEC plastic cased sockets and make a single ground point with as short a lead as possible to the chassis. The 2BA bolt used is long enough to accommodate any solder tags necessary but usually I ground the copper bus rail at this point only then run it on insulated stand-offs to where it is needed.

[Boulevardier]

"One of the things that crossed my mind as to the possible cause of instability was the lack of a grounded metal chassis."

I've been following this thread, and I'd also wondered about that very basic point. In the Mullard design, of course, the metal chassis is fundamental and will be key to all their performance figures. In the 5-10 that I built from a kit of parts many years ago, there were huge hum problems, and it was all down to an inadequate (metal) chassis and the way things were grounded thereon.

Mike

[Diabolical Artificer]

The chassis ground and mains earth are two separate things. I've had issues on amp builds where the chassis wasn't grounded, EG ground bus tied to chassis. If you think about it a big metal chassis can act like one plate of a capacitor so any metal parts, EG just about everything can then form part of the other plate.

Same goes for heaters in a way, heaters should not be floating so either ground one end of a winding or use two hundred - ish ohm R's to form a center tap, you probably know this anyway.

Your mains socket even if metal cased should be earthed I would have thought, but worth checking, if you then fit a earth tag to the chassis, you then have an earth loop. Looking at an IEC socket with metal shielding can and filter I have, the can is part of the IEC earth tag. Wiser heads than mine here will know whether this in a potential issue.

Here's a link to a thread over on DIYaudio that discussed possible issues with tag strips - https://www.diyaudio.com/forums/tube...reamp-hum.html and a link to another thread - https://www.diyaudio.com/forums/tube...db-proacs.html discussing "Mullard 5-20 topology" namely the Harmon Kardon Citation 5, though this uses completely different valves.

Andy.

[Kei-1986]

Originally Posted by Boulevardier

"One of the things that crossed my mind as to the possible cause of instability was the lack of a grounded metal chassis."

I've been following this thread, and I'd also wondered about that very basic point. In the Mullard design, of course, the metal chassis is fundamental and will be key to all their performance figures. In the 5-10 that I built from a kit of parts many years ago, there were huge hum problems, and it was all down to an inadequate (metal) chassis and the way things were grounded thereon.

Mike

Hopefully, since I didn't have any significant issues with noise, I don't think the problem was with the chassis material. It definitely did pickup external EM interference as it would amplify switching noise from my plasma TV and my pc even if it wasn't connected to the device. It just simply needed to be in the vicinity.

Originally Posted by Diabolical Artificer

The chassis ground and mains earth are two separate things. I've had issues on amp builds where the chassis wasn't grounded, EG ground bus tied to chassis. If you think about it a big metal chassis can act like one plate of a capacitor so any metal parts, EG just about everything can then form part of the other plate.

Same goes for heaters in a way, heaters should not be floating so either ground one end of a winding or use two hundred - ish ohm R's to form a center tap, you probably know this anyway.

Your mains socket even if metal cased should be earthed I would have thought, but worth checking, if you then fit a earth tag to the chassis, you then have an earth loop. Looking at an IEC socket with metal shielding can and filter I have, the can is part of the IEC earth tag. Wiser heads than mine here will know whether this in a potential issue.

Here's a link to a thread over on DIYaudio that discussed possible issues with tag strips - https://www.diyaudio.com/forums/tube...reamp-hum.html and a link to another thread - https://www.diyaudio.com/forums/tube...db-proacs.html discussing "Mullard 5-20 topology" namely the Harmon Kardon Citation 5, though this uses completely different valves.

Andy.

Grounding wise, I think I've got it sussed. With the tag strips that I'm using, I'm not using the tags that bolt to the chassis. They will be left unused, which should avoid accidentally creating any ground loops.

Eli did try to convince me to build a Citation V clone early on when I was planning which design to build as I was also considering a Pete millett uniamp. The mullard won out as it was very well documented and there were transformers available off the shelf.

[Herald1360]

Connecting safety ground (mains earth) to chassis close to any IEC inlet (using a dedicated screw and appropriate star washer) should be done anyway. It's a safety issue and shouldn't be negotiable. After that the signal and power supply 0V should have their own "trees" grounded to chassis at a single star point. Ideally no other connections to chassis should exist though you may find by experiment that you can "get away" with the odd less than ideal arrangement.

[trickie_dickie]

This is mine.

A long way to go. Haven't started wiring yet. All the transformers are from VVT. I'm adding negative bias via an EB91 and a magic eye just for show.
It has two Tagboards, one for the amp and one for the power supply. The Valves are Russian.

[Kei-1986]

Originally Posted by Herald1360

Connecting safety ground (mains earth) to chassis close to any IEC inlet (using a dedicated screw and appropriate star washer) should be done anyway. It's a safety issue and shouldn't be negotiable. After that the signal and power supply 0V should have their own "trees" grounded to chassis at a single star point. Ideally no other connections to chassis should exist though you may find by experiment that you can "get away" with the odd less than ideal arrangement.

That's what strikes me as odd is the mullard book literally states that the only physical link to the chassis is made at the input socket. The mains earth links to the bus bar at the edge of the tag board and follows the busbar to the chassis at the input socket.

Originally Posted by trickie_dickie

This is mine.
Attachment 204148
A long way to go. Haven't started wiring yet. All the transformers are from VVT. I'm adding negative bias via an EB91 and a magic eye just for show.
It has two Tagboards, one for the amp and one for the power supply. The Valves are Russian.

That looks rather good. Is that built on a hammond chassis?

Progress so far has been fairly good. Heater wiring is proving a little bit more challenging with the standard layout. I'll also have the challenge of soldering the live/neutral wires onto the on/off switch, making sure they are sleeved and remain twisted.
[img]https://***********************/65535/49830748643_42fb501db1_h.jpg[/img]

[Boulevardier]

Originally Posted by Herald1360

Connecting safety ground (mains earth) to chassis close to any IEC inlet (using a dedicated screw and appropriate star washer) should be done anyway. It's a safety issue and shouldn't be negotiable. After that the signal and power supply 0V should have their own "trees" grounded to chassis at a single star point. Ideally no other connections to chassis should exist though you may find by experiment that you can "get away" with the odd less than ideal arrangement.

But where should you connect the mains safety earth if, like the OP, you're using a wooden chassis?

Mike

[Diabolical Artificer]

Originally Posted by Boulevardier

But where should you connect the mains safety earth if, like the OP, you're using a wooden chassis?

Any exposed metal parts probably like tfmrs, however I've just read somewhere that OPT's should be isolated from the chassis (being a normal metal type).

Andy.

[rontech]

Originally Posted by Boulevardier

But where should you connect the mains safety earth if, like the OP, you're using a wooden chassis?

I imagine you would run a busbar around the appropriate areas of the "chassis" and connect the mains to the star point at the earthy end of the busbar

[Kei-1986]

Whilst putting this together, I've noticed that all of the 1M resistors I've got seem to measure around 600K with the meter wandering between 450K-700K. It seemed really odd, so I pulled out a second DMM and that did exactly the same thing. I think I'm going to have a dig in my resistor box and find some that measure properly. Glad I've been measuring each component before it goes in.

[TonyDuell]

You weren't touching the metal parts of the probes or the resistor leads when you were measuring them I assume? If you were then your body resistance ends up in parallel with the resistor under test and for something like 1M it will make quite a difference.

[Sideband]

Indeed it will. Depending on how moist your hands are your resistance could be anything from 100K to 500K and varying....try holding the test prods and checking.

[Kei-1986]

Originally Posted by TonyDuell

You weren't touching the metal parts of the probes or the resistor leads when you were measuring them I assume? If you were then your body resistance ends up in parallel with the resistor under test and for something like 1M it will make quite a difference.

yep. Some days I do the daftest things without realising. Thank you for pointing that one out, definitely one to remember for the future.

Made good progress today. I've only got the main ground bus to finish along with some cable tying. The floating resistors off the EL34's will go to the ground bus.
[img]https://***********************/65535/49835341027_7994c5676e_h.jpg[/img]

One thing I have noticed is that in the standard mullard layout, the connection for the feedback is remarkably long and twisted with the output transformer common

[Diabolical Artificer]

Ooooh, been caught by that one, thing is your clips never clip R leads properly and you never have enough hands, I'm thinking of getting a trained octopus.

That looks neat, htr wires shoved right in the corners, did spot one possible wiring issue, the heater wiring looped round the side of those two bases at the bottom middle - ECC83's ? to pin 9. See here, bottom page - http://www.valvewizard.co.uk/heater.html . Perhaps leave it for now, de-soldering it will knacker the wiring, if you have an issue bear it in mind.

"One thing I have noticed is that in the standard mullard layout, the connection for the feedback is remarkably long" Shouldn't be an issue, but best to keep short as poss like all wiring to cut down on stray C, especially wiring inside the feedback loop. Might be an idea to temporarally fit the FB R&C, you might need to tweak the C especially, when square wave testing. Your OPT's LC will affect things somewhat.

Andy.

[Kei-1986]

Finished and tested. Found one wiring error which caused oscillation on the first test. I got the wiring from C10/C11 backwards. Other than that all seems ok. Testing off the signal generator shows very similar behaviour to the MK1. Same ringing on square waves and on overload. I've noticed that as signal level increases in the 10KHz+ region, the phase shift between input and output seems to rise. That will need further testing, but it's oblivious that the feedback will require some adjustment.
[img]https://***********************/65535/49838552048_fa0a20d439_h.jpg[/img]

Some other problems, I noticed are that there is considerable hum when connected to my preamp which the mk1 didn't exhibit. It's silent with no input, which suggests a ground loop. I also noticed during testing, one of the output valves was red plating again. This was with the 60 ohm dropping resistors in place and the JJ GZ34 rather than the sovtek 5AR4 which is supposed to be "saggier". The rosy glow didn't make much sense as the voltages I measured all checked out ok though they were measured about an hour earlier. (433V on the anodes and grid, ~ 30.5V on the cathodes)
[img]https://***********************/65535/49839090421_3d6757c2d3_h.jpg[/img]

Primary differences at the input are the length of the input resistor wire is ~ 2 inches and I forgot to put the isolating washer in the input socket that isolates the ground from the body of the socket. (not that I expect that to make any difference)

[Diabolical Artificer]

"I got the wiring from C10/C11 backwards" easily done. "Same ringing on square waves and on overload" perhaps try a 220r stopper on the screen grids, surprised Mullard didn't fit em, still, their test OPT may not have needed it. Failing that up g1 stopper to 2k7, 270r on g2's. You may even have to up them further still, however the EL34 intergrid Miller C isn't big compared to other power valves, still, worth bearing in mind. Playing with the values of C1 and R3 may help too as might triode strapping the EF86 or reducing R6, known 5-20 issues.

"Some other problems, I noticed are that there is considerable hum when connected to my preamp which the mk1 didn't exhibit" Typical, some sniffing abouts needed there.

" I also noticed during testing, one of the output valves was red plating again." What DMM do you have, has it a good frequency range? Might be worth popping it on the OP on AC V and seeing if it reads anything without an IP, same goes for scoping the OP, all indications are that the amp is oscillating, or one valve is, have you any other EL34's you can try?

"The rosy glow didn't make much sense as the voltages I measured all checked out ok" Flick the meter onto AC after doing DC condition checks, it's often illuminating.

The open loop gain of the standard 5-20 has oodles of gain, I'm sure more is needed on the front end to roll off HF gain, a better LPF between the EF86 & ECC83. Anyroad, hope you get it sorted, it's a bummer when you make changes, put in a lot of work only to have nothing much change, however this reveals your layout on MK1 was pretty much ok, so, not a layout problem on the whole.

Andy.

[Kei-1986]

Originally Posted by Diabolical Artificer

"I got the wiring from C10/C11 backwards" easily done. "Same ringing on square waves and on overload" perhaps try a 220r stopper on the screen grids, surprised Mullard didn't fit em, still, their test OPT may not have needed it. Failing that up g1 stopper to 2k7, 270r on g2's. You may even have to up them further still, however the EL34 intergrid Miller C isn't big compared to other power valves, still, worth bearing in mind. Playing with the values of C1 and R3 may help too as might triode strapping the EF86 or reducing R6, known 5-20 issues.

The one improvement that was evident was the low frequency stability. Testing at 10Hz didn't see it flipping into HF oscillation like it did on the MK1. That seems to be the only improvement I've observed.

Based on this schematic (for component numbering continuity), aren't the R24/R25 1K resistors the screen grid resistors? R18/R19 2.2K are the G1 stoppers.
[img]https://***********************/65535/49573599056_daf4e1200b_h.jpg[/img]

Triode strapping the EF86 was at the top of my list of modifications to consider, however I've got no idea on what other changes would need to be made beyond strapping pin 1 to pin 6. One thing I have just noticed which may or may not have an effect is pins 2 & 7 on EF86 are tied together on the socket and that's it, I never grounded them on the MK1 or MK2. I assume these connect to the shielding structure within the valve and is likely to simply affect noise rejection.

Originally Posted by Diabolical Artificer

"Some other problems, I noticed are that there is considerable hum when connected to my preamp which the mk1 didn't exhibit" Typical, some sniffing abouts needed there.

"I also noticed during testing, one of the output valves was red plating again." What DMM do you have, has it a good frequency range? Might be worth popping it on the OP on AC V and seeing if it reads anything without an IP, same goes for scoping the OP, all indications are that the amp is oscillating, or one valve is, have you any other EL34's you can try?

"The rosy glow didn't make much sense as the voltages I measured all checked out ok" Flick the meter onto AC after doing DC condition checks, it's often illuminating.

I use a fluke 28II which seems to have a decent frequency counter for AC. I've got a total of 6 output valves to play with. 4x EL34B's and a pair of KT77's.