Bogen CHB-50 Rebuild

[physics]

Hi all!

As mentioned in the CHB-20 rebuild thread, my friend decided that 50W was a better fit for his needs so I am working on making a Bogen CHB-50 usable as a clean pedal platform. Schematic for the curious:
   

My primary mission right now is rebuilding the power supply with some new electrolytics and also a film cap to replace the paper-in-oil one from the bias supply. However, there are a few other caps (coupling caps) which I'm unsure on whether to replace or not. I think they are film types since they don't look like paper-in-oil caps that I have seen before, but I was hoping some of you would recognize them. I've heard chatter on other forums that old coupling caps between stages should be replaced too even if film types, but given the same forums also worship at the altar of all sorts of audiophoolery that doesn't seem like a reliable source of advice. Are there any guidelines for determining whether coupling caps are bad, assuming one doesn't have a fancy high voltage capacitor checker?

Pictures of the caps in question:

   
   
   

Thanks for any input!

[Sherlok Ohms]

Hello Physics

That is an interesting circuit which should be very pedal friendly as is. I would suggest that you simply get it working first and have your friend assess the tone. It should be very "high-fidelity" as it was intended to be a clean public address amplifier, and not anything approaching the kind of tone altering beast modern guitar players use.

The capacitive-coupled bias supply alarms me. I am in agreement with KOC that such bias supplies should be replaced with proper transformer-generated supplies, and a true independent bias-set network applied with cathode current-sense resistors for the output valves. At the moment there is a mix of fixed-biasing with cathode biasing. This is not bad unto itself, and if I recall, Mr. O'Connor did that in previous versions of his Studio power amplifier.

Regarding the capacitors: I think it is safest to replace the ones that say "PMK" on them as the "P" may mean paper? I do not know this as fact, but the age of the unit suggests it could be true. All paper dielectric capacitors have a limited life regardless of what oil is used or how well sealed the package may seem. In fact, both the oil and the paper are organic and cannot have the durability and stability that plastic has.

This is obviously a hand-built amplifier with large solder connections that themselves are quite durable over time. However, as experts have warned us, solder is eroded by electron flow, so you may want to check certain connections that seem dubious.

Also, for completeness, you may want to replace 20% tolerance resistors with 5% or 10% as the former are cracked carbon and the latter would be carbon film. All carbon resistors exhibit distortion unique to carbon, but the earlier construction forms were quite poor and lacked stability, and lacked even the accuracy of manufacture one expects today. These deviations introduce nuisance noises, snaps, and pops, that would simply be annoying especially after investing time and effort to restore an interesting amplifier.

Cheerio

[physics]

Hello Physics

That is an interesting circuit which should be very pedal friendly as is. I would suggest that you simply get it working first and have your friend assess the tone. It should be very "high-fidelity" as it was intended to be a clean public address amplifier, and not anything approaching the kind of tone altering beast modern guitar players use.

The capacitive-coupled bias supply alarms me. I am in agreement with KOC that such bias supplies should be replaced with proper transformer-generated supplies, and a true independent bias-set network applied with cathode current-sense resistors for the output valves. At the moment there is a mix of fixed-biasing with cathode biasing. This is not bad unto itself, and if I recall, Mr. O'Connor did that in previous versions of his Studio power amplifier.

Regarding the capacitors: I think it is safest to replace the ones that say "PMK" on them as the "P" may mean paper? I do not know this as fact, but the age of the unit suggests it could be true. All paper dielectric capacitors have a limited life regardless of what oil is used or how well sealed the package may seem. In fact, both the oil and the paper are organic and cannot have the durability and stability that plastic has.

This is obviously a hand-built amplifier with large solder connections that themselves are quite durable over time. However, as experts have warned us, solder is eroded by electron flow, so you may want to check certain connections that seem dubious.

Also, for completeness, you may want to replace 20% tolerance resistors with 5% or 10% as the former are cracked carbon and the latter would be carbon film. All carbon resistors exhibit distortion unique to carbon, but the earlier construction forms were quite poor and lacked stability, and lacked even the accuracy of manufacture one expects today. These deviations introduce nuisance noises, snaps, and pops, that would simply be annoying especially after investing time and effort to restore an interesting amplifier.

Cheerio

Thanks for the feedback! Would you suggest going into one of the mic inputs, or the auxiliary input? I had figured that the mic inputs would not be ideal for guitar given the grounded cathode (mu amp?) configuration and was figuring on either going into the aux inputs if I wanted to be quick or adding some cathode resistors to the mic input stages if I wanted to go through there.

Regarding the bias supply, my tentative plan is to add per-tube adjustment and current sensing, however I have not given much thought to supply impedance beyond planning to bypass the wipers of the high-value pots (250k each) to ground with capacitors to keep a low-impedance grid leak circuit at signal frequencies as mentioned in TUT2. Now that you mention it, I'll have to look more closely how well the supply will actually do at supplying current. I'm trying to avoid non-essential costs, but then a 115V to 6.3V chassis mount hammond transformer is only around $10 plus a few extra bucks for lower value pots, and I think I have the rest of the parts lying around. RBX + BMK is probably out of the budget for this (around 60USD) otherwise I'd likely just go with that instead of sourcing parts myself.

Yeah, definitely hand built. Some stuff is nicely done as well, such as local decoupling for the mic gain stages and (some) wired grounds. Other stuff less so, like multiple chassis ground points.

The age of the unit is definitely part of why I'm worried they may be paper, however I'm trying to find something definitive along those lines to avoid adding a possibly unnecessary cost. If I can't find anything though, then yeah definitely agree it's safer (and not that costly) to just replace them with some plastic caps.

I'll keep the resistors in mind for sure, thanks for mentioning that

Thanks!

[makinrose]

Nice amp! The I'd suggest keeping one of the channel as grid-leaked baised because it is interesting sound but the main channel should be converted to cathode-bias since the grid-leak biased inputs cannot tolerate really high input signals that some pedal players may use. The Aux inputs are meant as line ins so they may not be too useful.

These are not all that hi-fi. They have character. The overall build is rather tweed-like. For a more modern player you probably will need to roll off some low end to keep it from being woofy and there are places to add gain if you want to. You could also experiment with the negative feedback loop.

It looks like a fun project!

[physics]

Nice amp! The I'd suggest keeping one of the channel as grid-leaked baised because it is interesting sound but the main channel should be converted to cathode-bias since the grid-leak biased inputs cannot tolerate really high input signals that some pedal players may use.  The Aux inputs are meant as line ins so they may not be too useful.

These are not all that hi-fi.  They have character.  The overall build is rather tweed-like.  For a more modern player you probably will need to roll off some low end to keep it from being woofy and there are places to add gain if you want to.  You could also experiment with the negative feedback loop.

It looks like a fun project!

Thanks for the feedback! Especially regarding the grid-leaked bias stages. Was originally planning on just rewiring both of them, but if the sound is at all usable I'd say it's worth leaving at least one like that for my friend to try.

[physics]

The capacitive-coupled bias supply alarms me. I am in agreement with KOC that such bias supplies should be replaced with proper transformer-generated supplies, and a true independent bias-set network applied with cathode current-sense resistors for the output valves.

Could you expand on what alarms you? So far the issues I've found while digging around online/in TUT is that it may not be able to supply enough current in certain situations (like using BMK, or high grid drive) and that some amps have a reputation for blowing the coupling capacitor (Marshall something or other, with a plastic coupling cap). I agree that a transformer-based supply would be best, at the moment I'm trying to work out if it's necessary in my situation either for current supply capability or immediate danger (hence asking about why you're alarmed). Even if it does turn out that I don't need a higher current supply, is there any benefits towards building a higher current supply anyway, whether from a tone or robustness perspective?

Thanks for any info.

[physics]

Alright, been working on this under the radar for a bit, got a new power supply board almost finished for it. While doing that, I noticed that the schematic calls for a 10R power resistor from one end of the plate winding to the voltage doubler (see picture below). Anyone have any idea why that is there or if it's needed? Only things I've thought of so far is that it's to limit charging current and maybe to act with the cap(s) as an RC filter, maybe some sort of voltage division scheme?
   

Thanks for any input!

[K O'Connor]

Hi Guys

The 10R-10W is simply for current limiting into the filter caps.

Capacitively-coupled bias supplies are the poor/cheap/lazy man's way of making a negative rail without a proper winding available on the PT. In most cases, the coupling cap is under-value and the voltage rating is wrong - also too low. Such supplies are generally too low a voltage to provide proper control of the tubes particularly when the main plate supply uses the same winding with a full bridge.

Intuitively it seems like a worse proposition when the winding for a voltage doubled plate supply is used. But... in each case, the circuit would be impressively referred to as a "charge pump", which can be constructed in a reliable manner, but doing so makes the economics of it less impressive compared to adding an auxiliary PT given the small size of the latter.

For your interest: Doing the charge pump bias supply closer to correct requires using only polypropylene for the plastic coupling cap AND making sure the cap is properly rated to withstand the winding voltage. You may find that a 440VAC safety cap will suffice, and these are available up to 10uF.

Have fun

[physics]

Hi Guys

The 10R-10W is simply for current limiting into the filter caps.

Ok, thanks! I'm guessing it needs to stay in? I was hoping it was to limit current into subpar caps, and that I don't have to worry about that anymore since I'm installing some good modern capacitors, though I guess it could be possible other parts of the circuit depend on the current limiting too such as the diodes or transformer. Diodes are easy to check specs/replace, but if the transformer isn't up to the task that's a stickier problem and I'm not sure how I could test it to find out.

[K O'Connor]

Hi Guys

Okay... a physics question:

How much current will an ideal capacitor pull in at the initial application of voltage across it, assuming an ideal voltage source?

Hint: The answer is in TUT3.

[physics]

Basically infinite.

[K O'Connor]

Hi Guys

Don Rickles would say "You win a cookie"

So you can see that the 10R is pretty important in saving the life of the PT and the diodes and likely the caps. All of these components in the real world are far from ideal, but every bit helps.

[physics]

I was just gonna ask how come I don't see this resistor in other amps that often, but flipping through TUT5 I keep seeing a comparable resistor so it seems they make it into your amps at least. Both the Bogen CHB-20 and the CHB-100 amp schematics don't show a resistor like that though, which is what initially piqued my interest regarding possibly removing it to save board space. Now though I'm thinking it might be a good idea to add one to those other amps missing it.

Edit: In particular with TUT5, the stock Marshall Major supply doesn't have a limiting resistor from what I recall, but your re-worked ideal power supply for it does.

[K O'Connor]

Hi Guys

You will notice there are a lot of extra parts in the preferred schematics compared to OEM. Manufacturers want to save money and time, so removing a "nicety" saves the cost of that part, the cost of the labour to install it, and the time to install - maybe they no longer have to inventory it?

Choke-input supplies provide natural current limiting. Chokes were easier and cheaper to manufacture than were high-voltage filter caps of significant value, but then capacitor technology and economics changed and capacitive-input supplies became the norm.

You see current-limiting resistors in older circuits using tube rectifiers and old texts say to add them as do the tube data sheets. As consumer electronics evolved, these resistors were dropped for the above-mentioned reasons. In some amps where parallel low-voltage output stages are used - as in MI and most hifi - parallel rectifier tubes might be used, which effectively increases the current capability of the net rectifier.

Back when I knew less than I know now, I would look at these low-value Rs and Cs and think they are unimportant and exclude them from my rendering. But then you learn more and realise what those parts are for and add them back in. There are some circuit places where I add in a resistor, say, where I never see one in simplified app notes nor in complete sophisticated products. An example of this is a resistive tether to ground at the output of an opamp or direct-coupled solid-state power amp. To me this R helps with reducing DC offset by providing a leakage path for the output stage. It costs pennies. It may not actually do what I initially added it to do, but it does contribute to stability at very high frequencies.

[physics]

I guess I expected there to be the same cost-cutting measures across the whole product line, so it threw me for a loop when the 50W model had it but not the 20W or 100W models.

What are the design guidelines for choosing these resistors? Since my last message I've been working out that detail in parallel to getting the power supply rebuilt after switching to an eyelet card (the original salvaged turret board was too delicate, could barely handle heat). So far I've checked that the average power dissipated by the voltage doubler diodes due to ripple is within margins, and it looks like ambient temp can get up to 51 C -- 66 C before the 1n4007 diodes get above 60% -- 70% of max junction temp, so that should be fine I figure. The bias supply should be fine too since it's much lighter duty, but I haven't explicitly checked that.

I'm currently hung up on the diode surge current ratings right now. For the voltage doubler things look fine, I have two half-wave rectifiers stacked on top of each other and each one has a 100uF cap fed by a 10ohm 5W resistor (per original schematics) which means assuming 265Vpeak from the transformer Ipeak is 26.5A at turn on, which is less than the 1n4007's 30A peak surge into a resistive load. The charging waveform should be just a quarter sine instead of the half sine the peak current is given at too, so that's some extra margin. The bias supply is a different story though...

For the bias supply it's just a bridge rectifier of 1n4007's feeding a 100uF cap through a 1ohm 5W ohm resistor. Initially I thought I'd be fine with this since it's basically the same as the B+ for The Standard in TUT5 (1n4007's feeding 88uF through 1 ohm 5W resistor, with no powerscaling )and my peak VAC is a fraction of what The Standard's B+ supply takes.  Should add that I recently noticed that the cap bank in The Standard is actually fed through 2 x 1ohm 5W resistors plus another 1n4007, but the B+ input voltage is more than 2x that of the bias supply I'm putting together so that just makes them a little more comparable. For the bias supply I'm putting into this amp, I plan to run a 120VAC to 6VCT transformer in reverse off the heater winding, meaning I'll get about 90Vpeak out of it. At turn on then, peak current draw is 90A! Now, average current is less than that, but I know average current alone doesn't account for differences in waveform shape or the change in voltage drop with forward current. Due to that, I'm still not entirely sure how to work from the peak current waveform in the datasheet back to whatever waveform I have and make sure it's in spec. A rough method might be to pick the lowest voltage drop across the diode for the half cycle and multiply by the average current to get a lower bound on max average surge power.

After seeing that there'd be a giant spike at turn-on, I figured I'd double-check what the B+ supply looks like for The Standard at turn-on, since if that's more demanding and is fine design-wise, then my bias supply shouldn't be an issue. Anyhow, the 272JX transformer in the TUT5's The Standard is 600Vrms center-tapped, so when applied across a 2-ohm load at turn-on that's a peak of over 300A! Again average current is less over that conduction cycle, roughly 12.7A, but like I mentioned before I don't think I can go off average current based on what I think I understand so far. Obviously my little 90V bias supply should be fine if The Standard's B+ supply is fine, but I don't understand why it's fine.

I could always just chuck in some 1n5408 diodes into the bias supply instead if I didn't want to mess with either the cap size or RC constant (right now it's just barely under 1/10th of the B+'s RC constant), but it seems a bit ridiculous to have such beefy diodes in the bias supply while having 1n4007's in the voltage doubler B+ supply. Plus, I want to actually figure out what's going on here so that I'm better at electronics instead of just chucking bigger parts at circuits when I don't understand them, as valid an approach as that may be.

So, what am I missing? If anyone can direct me to some good books/resources on this particular aspect of power supply design, that'd be much appreciated too.

Thanks for any input!

[physics]

I'm seeing comments elsewhere on the interwebz about I^2 t ratings... I'm gonna go dig into that and see if that's the answer.

[physics]

Alright, new results, with calculations to come later. Short story is that one of my power supply design books states that for the characteristic in question, the Nonrepetitive Peak Surge Current, gives the acceptable RMS value (half of peak, in the case of the half-sine waveform used in datasheets) for the specified period of time. 1n4007's are rated for 30A peak, so 15A RMS over one cycle is the limit. I crunched the integral for RMS of a charging capacitor (assuming DC voltage source to charge, should be worst-case), plugged in my circuit values, and my bias supply's surge current is about 5A RMS. Well within 60% derating, so that should be fine. I'll post calculations 'n stuff later for those of you playing along at home, and in case I made a mistake somewhere.

[physics]

Aha! 600VCT RMS means 300V RMS from center-tap to outer-tap. Still get tripped up by that... Anyhow, plugging that into my derived formulas, The Standard's turn-on surge is roughly 15.4A RMS. I made some worst-case assumptions though like charging the caps from DC, so it could be a bit less than that. When I get a chance I might try to get a more exact calculation.

[K O'Connor]

Hi Guys

These extra resistances we add to limit surge currents are also there to enforce a high-frequency filter effect, but...

In the context of surge current limiting, you are forgetting the dynamic resistance of the diode itself, plus the PT winding resistance, plus the connecting wiring itself. All of this adds up to the fact that standard 1N4007s can be used almost universally in guitar amps <100W. For essentially no mark-up, you can use the faster UF4007. Or, if you stick with the 1N-series, use 1N4004s in series as those are as fast as the UFs.

If you are contemplating going to 3A diodes, think twice. They have huge fat leads since they are meant to cool the semiconductor die. You will find more amenable easy-to-solder and easy-to-place into eyelet leads on most integrated bridges. If you use a bridge with a CTed plate winding, be certain to add a resistor to ground off the "unused" negative terminal. Of course, there should be one off the positive terminal, as standard practice. These will prevent damage to bridge.

The voltage drop of a solid-state diode in a tube plate supply is immaterial and need not complicate the calculations. Besides, varying mains voltages and loading conditions will make an attempt at precision pointless. Another significant point is that there is only about a 15% chance of turning the amp 'on' at the mains peak voltage.

Have fun

[physics]

Hi Guys

These extra resistances we add to limit surge currents are also there to enforce a high-frequency filter effect, but...

In the context of surge current limiting, you are forgetting the dynamic resistance of the diode itself, plus the PT winding resistance, plus the connecting wiring itself. All of this adds up to the fact that standard 1N4007s can be used almost universally in guitar amps <100W.

Yep, that fixed the numbers. The bogen chb50 transformer's plate winding is about 7.5 ohm, taking that as the resistance of the 272JX plate winding and plugging that into the formula I get about 7A RMS for The Standard's surge current. Knew I must have been missing something. Winding resistance and inductance were left out originally to simplify things and get a worst-case value, but it looks like I got too far from reality. I hadn't thought about the diode's dynamic resistance or the wiring resistance, good point. Re-doing my calculations for the Bogen's surge with the plate winding resistance accounted for, it looks like I can add an extra 100u filter cap in parallel with the doubler and still keep nice comfy margins. Great! I had wanted to add extra filtering or have the option to, but up 'till now I thought I might overheat the diodes on startup.

Thanks for clearing that point up!

For essentially no mark-up, you can use the faster UF4007. Or, if you stick with the 1N-series, use 1N4004s in series as those are as fast as the UFs.

What's the benefit of faster diodes? Better efficiency of the supply?

If you are contemplating going to 3A diodes, think twice. They have huge fat leads since they are meant to cool the semiconductor die. You will find more amenable easy-to-solder and easy-to-place into eyelet leads on most integrated bridges.

Yup, that was my main hesitation with the 3A diodes. I can just barely cram two of their leads into the Keystone 44 eyelets, which hardly leaves any room for whatever else needs to go in.

If you use a bridge with a CTed plate winding, be certain to add a resistor to ground off the "unused" negative terminal. Of course, there should be one off the positive terminal, as standard practice. These will prevent damage to bridge.

I assume by "negative terminal" you mean the unused outer tap on the transformer, assuming I hook up the bridge between the center tap and one outer tap? How do these help protect the bridge? As a guess, maybe there are inductive spikes associated with diodes turning off and on, and the resistors help damp that?

Your comment about transformer winding resistances unearthed a few other problems with my bias supply design. After you mentioned the importance of transformer winding resistance as it relates to surge current, I went and measured the transformer I was planning on using for the bias supply. The line-side winding that I was going to use as the high-voltage source for the bias supply is 126 ohms! So my RC constant for the bias supply is an order of magnitude higher than the B+'s RC constant, and high-voltage will be applied to the tubes before they're properly biased. That's inconvenient. So looks like I either need to find a different transformer or maybe work out a soft-start circuit.

The other problem is that the line-side of the transformer is rated for 80mA max. Fine during normal use of the bias supply, however at turn-on I calculate around 200mA RMS leaving out the diode dynamic resistance and inter-component winding, but including the transformer winding resistance. Is this going to be an issue? I'll dig around in my books and the internet to see what I can find quantitatively, but I figured I'd ask.

The voltage drop of a solid-state diode in a tube plate supply is immaterial and need not complicate the calculations. Besides, varying mains voltages and loading conditions will make an attempt at precision pointless. Another significant point is that there is only about a 15% chance of turning the amp 'on' at the mains peak voltage.

Right, precision is a losing battle. I'm instead trying to get reasonable upper bounds on things so that I know in the worst-case I should be fine. I figured there's a small chance of turning on the mains at/near peak, but it seemed large enough that it'd happen more than once in the lifetime of the amp so I didn't think I could ignore it.

Thanks for your input! It definitely helped clear up stuff with ratings, though I suppose now I have new problems to deal with. The hamster wheel spins on...