Bogen CHB-50 Rebuild

[K O'Connor]

Hi Guys

With a traditional CTed plate winding on a PT designed to support a tube amp, there are two options when using an integrated bridge rectifier.

The first option is basic and the AC terminals of the bridge merely tie to the ends of the winding. The bridge has to be rated for the full winding voltage. For example, if the winding is 300-0-300Vac, that is 600Vac total which produces a peak voltage of 848V. The rated voltage is at full load and the regulation rating for the PT will suggest what the unloaded voltage will be, where lower-VA devices have worse regulation and their unloaded voltage will be higher. Say it is 20%, then the unloaded peak is now just over 1kV - 1,060V - so the bridge must be rated for >1kV.

In this example, because the CT is tied to ground, the transformer produces +/-420Vdc (loaded; up to 470Vdc unloaded0, once filter caps are added, with the usual bleeder resistors to ground from each DC output. Of course, you are unlikely to need such a high negative voltage although you now have full-wave pulsating DC to generate a bias voltage from.

Further with this same example, if there is a bias tap on the winding, it will still look like its AC value, usually around 50Vac, and can be half-wave rectified for bias or other uses.

In a cathode-biased amp, the negative end of the bridge simply ties to ground through a bleeder resistor to protect the bridge. No cap.

The alternate bridge wiring would be for the bridge to span from the CT to one end of the winding. The free winding end is insulated and stored. The DC output of the bridge is handled in the usual way with a cap and bleeder resistor. In the case of a plate supply, the negative end of this supply ties to ground. Bias should be derived from a separate PT if needed, or the winding and bridge can be rewired as above.

With regard to an aux winding or PT generating bias: There is no worry about which voltage rises faster. None of the voltages used in typical guitar amps are high enough to cause cathode stripping, as TUTs state, so this is a non-worry. Even were B+ to rise ahead of C- the tubes are not warmed enough for current to flow. Usually the supplies rise at about the same rate despite their seemingly different impedances.

The reverse-wired auxiliary PT for bias is usually low-VA which results in a voltage loss of twice the regulation rating. Tiny PTs are a compromise of design and even when used with forward connections, the loss is significant and certain adaptations are made to get useful throughput of energy. The reverse connection is more lossy and is very useful in some cases to achieve a more controlled output - another topic. In the bias supply context, it has sufficient output and remaining VA capacity to do the job well.

I have expanded on this and made two new threads in the Power Supply Design section.

[physics]

Hi Guys

With a traditional CTed plate winding on a PT designed to support a tube amp, there are two options when using an integrated bridge rectifier.

The first option is basic and the AC terminals of the bridge merely tie to the ends of the winding. The bridge has to be rated for the full winding voltage. For example, if the winding is 300-0-300Vac, that is 600Vac total which produces a peak voltage of 848V. The rated voltage is at full load and the regulation rating for the PT will suggest what the unloaded voltage will be, where lower-VA devices have worse regulation and their unloaded voltage will be higher. Say it is 20%, then the unloaded peak is now just over 1kV - 1,060V - so the bridge must be rated for >1kV.

In this example, because the CT is tied to ground, the transformer produces +/-420Vdc (loaded; up to 470Vdc unloaded0, once filter caps are added, with the usual bleeder resistors to ground from each DC output. Of course, you are unlikely to need such a high negative voltage although you now have full-wave pulsating DC to generate a bias voltage from.

Further with this same example, if there is a bias tap on the winding, it will still look like its AC value, usually around 50Vac, and can be half-wave rectified for bias or other uses.

In a cathode-biased amp, the negative end of the bridge simply ties to ground through a bleeder resistor to protect the bridge. No cap.

The alternate bridge wiring would be for the bridge to span from the CT to one end of the winding. The free winding end is insulated and stored. The DC output of the bridge is handled in the usual way with a cap and bleeder resistor. In the case of a plate supply, the negative end of this supply ties to ground. Bias should be derived from a separate PT if needed, or the winding and bridge can be rewired as above.

With regard to an aux winding or PT generating bias: There is no worry about which voltage rises faster. None of the voltages used in typical guitar amps are high enough to cause cathode stripping, as TUTs state, so this is a non-worry. Even were B+ to rise ahead of C- the tubes are not warmed enough for current to flow. Usually the supplies rise at about the same rate despite their seemingly different impedances.

The reverse-wired auxiliary PT for bias is usually low-VA which results in a voltage loss of twice the regulation rating. Tiny PTs are a compromise of design and even when used with forward connections, the loss is significant and certain adaptations are made to get useful throughput of energy. The reverse connection is more lossy and is very useful in some cases to achieve a more controlled output - another topic. In the bias supply context, it has sufficient output and remaining VA capacity to do the job well.

I have expanded on this and made two new threads in the Power Supply Design section.

Thanks for the responses here and in the other mentioned section! After reading your response I realized that I forgot to factor in the rest of the RC network for the plate supply, the bits for the screen, PI, and the pre-amp nodes. I added those to the simulation (I've been using https://falstad.com/circuit/ to double check calculations and expectations, to make sure I'm roughly on target, to the extent that the simulation model is), and indeed the rise time for the plate supply ended up being about the same as the bias supply despite the plate node in isolation having a much shorter time constant. I assume the rest of the network outside the plate supply is why you said they rise at about the rate same usually? If so it would seem like another case of me simplifying more than I should.

My next focus with the amp is the bias set pots. I originally had some 250k pots I was planning on using, with some capacitors to drop from the wipers to ground to make a low-impedance AC ground. I had grabbed 250k because I thought that I saw 220k pots used like this in one of the TUTs as an example of modifying a bias supply, but I'm having a hard time finding that figure again. Maybe I misread. Anyhow, doing the math now, that configuration would give me an RC of a couple seconds for the caps on the wipers, which seems out of whack, so I'm working out the details for using lower value pots, like what I see in the TUTs, in the ballpark of 25k. Currently sorting through the output impedance of the bias supply and grid circuit resistance, I'll create a thread in the TUT FAQ section for my question related to that.

Getting close to a complete power supply!

[physics]

Alright, had a chance to work on this again. Since the last post, I've got the amp working.

I internally jumpered the two mic inputs for more gain / as a step towards Plexi-style mods in the future, installed a low-impedance bias supply with per-tube adjustments and external metering and adjustment, added an extra filter cap at the plate node in parallel to the voltage doubler, Implemented a galactic ground scheme the best I could without making significant changes to terminal strip connection layouts, added 1k screen resistors, reduced grid leak resistors to accommodate grid stoppers, and clipped out irrelevant circuit portions like the tape booster circuit. Aside from that things are basically stock. So I guess the pre-amp is stock, and the power supply and power amp are largely rebuilt.


The good

It works! I can plug in my guitar and make guitar noises! It's a pretty clean amp, and has some pretty cleans (IMO). With the volume on 2 it's pushing the limits of neighborliness for city living and pressurizes the air around you. It feels nice to play, with firm bass and a sound that fills the room through my detuned 2x12. Through non-worn speakers and/or cranked, I bet it can push a lot of air. Hopefully I'll get the chance to blast it somewhere more noise-tolerant soon. I need to play-test it anyway. I'm officially jealous of my friend's nearly-complete amp, and I'm motivated to convert my 100W Bogen given how much I like playing the 50W. It also looks like the 50W doesn't suffer from the voltage doubling distortion cathodynes can have based on my scope display when clipping the snot out of the power section, so that's less stuff for me to fix.

The bad

There are a few gremlins to work out still. Also, at the request of my friend, corners were cut in lead dress and ease of repair in favor of finishing quickly, so it's not very tidy inside as I was mostly laying things out on the fly and so working on it takes patience and a steady hand.


The primary issue right now is that it oscillates. I thought it might be lead dress, given I was sloppy near the end as I started running out of time before leaving the state, but it actually seems to be some sort of coupling between the guitar and the amp. The conditions it oscillated under are as follows:

• Amp hooked up to resistive load (4 ohm 100W Dale power resistor) and the impedance set to match.
  
• Master volume maxed.
  
• Treble maxed
  
• At least one input volume maxed
  
• Guitar close enough to the amp, and with the volume and tone pots on full and the guitar properly oriented.
  
• Alternatively, unplug the cable from the guitar and feel around with the now-free end near the power tubes and output transformer until finding a location and orientation that causes a squeal.
  

With the above configuration, I get an oscillation in the low kHz range, maybe around 2kHz-4kHz judging by my scope display. Things that change it:

• Moving the guitar away from the amp damps then kills the oscillation.
  
• Reducing the master volume kills it.
  
• Reducing the treble kills it.
  
• Certain settings of the bass knob might kill it?
  
• Guitar orientation can make or break the oscillation, even when close to the amp.
  
• Rolling down the volume knob (and/or tone?) on the guitar seems to damp then kill the oscillation.
  
• With the guitar volume all the way off, it oscillates again but at a lower frequency. So it first gets killed, then comes back lower.
  
• At one point, the position of my pickup selector seemed to affect it. Middle selection on my HH superstrat oscillated, but flipping to neck or bridge pickup killed it.
  
• Touching a control knob or the chassis may have affected the oscillation or made it easier to get. I'm not sure this wasn't just the guitar orientation changing.
  

Some other observations:

• Flipping the chassis over so that the tubes and transformer were on the other side relative to the guitar didn't stop me from getting it to squeal.
  
• With the chassis flipped, using a chopstick to poke around inside and move wires did not stop it from squealing or have any noticeable effect. Perhaps I didn't move the right wires.
  
• Putting the chassis into the metal head shell didn't help with the oscillation, except by preventing me from getting the unplugged jack close enough to the power tubes to start. The guitar still oscillated without issue, and it may have been easier to get the oscillation going too, counterintuitively.
  
• The tubes installed are an old 6Eu7, a new production 12AX7, and old 6C4, and new-production but fairly worn 6L6GC's from Ruby.
  
• The original 12AX7 functioned, but was microphonic. I could hit it, and it was like a bell through the speakers. I think I was hearing feedback through it at one point too. The new production one is much better. The 6EU7 was slightly microphonic, but not near as much and I don't have a replacement so it stays in for now. The 6C4 does nothing of note.
  
• The 6L6 tubes have a faint blue glow to them in certain areas. With the output power sufficiently high, playing will modulate the blue glow. Mostly off/on without any gradations.
  
• At one point it seemed like I could get a lower frequency oscillation by maxing the bass control, turning the treble off, and maxing the master volume, but I haven't been able to reproduce that and I'm not sure I didn't just touch the end of the input cable and inject interference. It sounded buzzy and almost arc-like the one time I got it to happen, which is pretty similar to how it sounds when I touch the end of the cable.
  
• I don't see any suspicious out-of-audio-range oscillations on my scope when the amp isn't squealing, nor when it is.
  
• I do see what appears to be a mains harmonic(?), 60kHz, at a very low level when all the controls are zeroed, and what appears to be diode rectification noise occurring every 8 ms or so.
  
• I've commented on how things sound, despite the amp being plugged into a dummy load. Something in the amp itself (transformer?) is apparently vibrating to make the noise I'm hearing. E.g., I can hear the oscillation despite there being no speaker connected to the amp.
  

So...

Any ideas what might be causing the sensitivity to guitar proximity? I've heard about this sort of stuff happening before, and the friend I'm working on this amp for actually has something similar happen with his EVH 5150 III EL34 50W. If he gets his guitar close enough (and with the gain/volume high enough?) he gets some awful sounding shrieks out of the amp. So maybe this isn't that unusual? "Other amps do it" doesn't seem like a good standard to go by though, especially when certain other amps can be played like a theramin. Also given the EVH has something on the order of 6-7 gain stages and a less than stellar engineering reputation, it's not exactly a role model.

Thanks for any ideas you guys might have.

[K O'Connor]

Hi Guys

Oscillation is generally caused by poor layout and lead dress. The original Bogen circuit lacks grid-stops, screen-stops and sufficient filtering, then incorporates "bandaids" to stabilise the circuit - low-value caps across plate resistors and other roll-offs to tame the circuit. The engineering ideal does not require any of these, but reality says they are needed.

The grid-stop values you have added may not be sufficient for the specific conditions within this chassis. There is no need to reduce the grid-leak value when adding a grid-stop, particularly with preamp tubes.

"Cathodyne voltage doubling distortion"?

Microphonic tubes can be tamed using an O-ring or a simple elastic band placed in line with the internal mica supports. This will at least last ling enough for you to determine if the tube status is significant to the oscillation problem.

The Bogen schematic shows some signals wired to the terminal strip which may become problematic for the guitar amp application. These extra paths may allow signals to interact that should not, and which may lead to oscillation.

Yes, the OT will physically vibrate with the signal and you can hear it when using a bench load. This is caused by magnetic leakage due to loose windings and imperfect magnetic shielding of the core - something that is always imperfect with non-toroidals.

Doing "as good of a Galactic ground as the terminal strips allow" does not sound like an actual Galactic Ground was achieved. A major part of this is how the filter caps tie in to the circuit. It may be impractical to achieve a straight-line signal path in some situations and this may require circuit amendments if gain has been increased from stock. Note that in the case of older amplifiers repurposed for instrument use, the way you use the amp is out of spec for the original circuit inasmuch as the intent was to run the amp clean, well below clipping. Running stages beyond their linear range may cause oscillations even without cascading things that were not previously cascaded.

[physics]

Thanks for the reply and info!

Oscillation is generally caused by poor layout and lead dress. The original Bogen circuit lacks grid-stops, screen-stops and sufficient filtering, then incorporates "bandaids" to stabilise the circuit - low-value caps across plate resistors and other roll-offs to tame the circuit. The engineering ideal does not require any of these, but reality says they are needed.

This was my first thought, however I don't know how to reconcile that with the oscillation only starting when I bring the guitar close to the amplifier. The fact that it goes away when the guitar/input end of the instrument cable is far enough away makes me wonder if the lead dress is fine (enough) to not be what's closing the feedback loop. Perhaps my mental model of oscillation is overly simplistic though.


Moving wires around inside the amp while it oscillated didn't seem to change anything, would that suggest anything? Perhaps that it's a layout issue not a lead dress issue?

The grid-stop values you have added may not be sufficient for the specific conditions within this chassis. There is no need to reduce the grid-leak value when adding a grid-stop, particularly with preamp tubes.

That's a good point, they're only 1k, and only on the power tubes. They were actually put in by a previous owner, and I just left 'em in. I should see what TUT says about suggested values. With regards to the grid leaks, they were at the max allowed for 6L6 already, 100k, so the 1k just breaks the max allowed bias circuit impedance. From what I had read in the TUTs it seemed like that was an important spec to observe, so I put in 50k resistors instead, since those are the parts I had with suitable specs. If I calculated it right, that'll lower the output of the amp but not significantly. Depending on the tube, I think I found that biasing at 50% max dissipation would still give 50W or just under. Are you saying I should have just left the grid leaks at 100k, and have a 101k bias circuit impedance for the 6L6 tubes? If so I guess that's another tube spec to take with a grain of salt.

"Cathodyne voltage doubling distortion"?

There is probably a better technical term for it. Merlin Blencowe talks about it on his ValveWizard site. He says that if the cathodyne inverter tube gets overdriven to the point of grid current, then the anode is not able to track with the cathode's current increase and you get part of the cathode output signal added to the anode's output signal. This results in a heavily distorted waveform with double the frequency of the input, roughly speaking.


Page where he talks about it: https://www.valvewizard.co.uk/cathodyne.html


Key picture (bottom one):

Microphonic tubes can be tamed using an O-ring or a simple elastic band placed in line with the internal mica supports. This will at least last ling enough for you to determine if the tube status is significant to the oscillation problem.

I had thought those were snake-oil as I don't see how they'd do anything to affect the internals besides just adding more mass to the glass envelope and changing how it vibrates slightly, but if you say it helps than I'll give it a shot. O-rings are cheap anyhow.

The Bogen schematic shows some signals wired to the terminal strip which may become problematic for the guitar amp application. These extra paths may allow signals to interact that should not, and which may lead to oscillation.

Assuming you're talking about the Remote #1 & #2, the Hi-Z input and the Tape Booster input, those and the associated circuitry have all been removed and those terminals grounded. I also disconnected the 70V transformer lead and heat-shrinked the end. I will double check if any of my new wiring on those strips brings other signals nearby. I think I'm good, but I may have grounded the unused terminals with the input signal's ground point not the speaker output jack's ground.

Doing "as good of a Galactic ground as the terminal strips allow" does not sound like an actual Galactic Ground was achieved. A major part of this is how the filter caps tie in to the circuit. It may be impractical to achieve a straight-line signal path in some situations and this may require circuit amendments if gain has been increased from stock.

Right, I should clarify. I believe a followed everything prescribed for the galactic ground scheme, however the Bogen was layed out and designed to have V1A, V1B, and V2A all fed from the same supply node. I did not have the time to cook up another node and make the necessary layout and circuit modifications, so those three triodes all share the same star. Thinking about it, I think that actually fell under interstage decoupling, not galactic grounds in the TUT series, but I might be wrong.

Note that in the case of older amplifiers repurposed for instrument use, the way you use the amp is out of spec for the original circuit inasmuch as the intent was to run the amp clean, well below clipping. Running stages beyond their linear range may cause oscillations even without cascading things that were not previously cascaded.

That's a good point I hadn't thought about. I figured that as long as my lead dress was proper and I didn't screw with the time constants too much then it should still be stable if it was originally stable.

[K O'Connor]

Hi guys

As TUT2 reported, low-frequency oscillations can be set off by high-frequency oscillations, but either is the result of too much gain between two in-phase signal points. Some of the low-frequency gain should be sacrificed for stability, by reducing some plate/grid coupling cap values: First to attack are to the 6L6 grids, reduce to 22nF each; remove the cap to ground off the splitter grid; remove the bypass across the feedback resistor; remove the plate bypass across the 470k for V2B.

Is the 150k-2W tied V2B cathode still in place? I would remove it.

The feedback loop could be opened, or altered depending on how other modifications go.

Since there are no grid-stops except for the 1k for each 6L6, you need to add a few. Use at least 68k for V1A and V1B although I would use 100k or higher. Same for V2A. Personally, I do not like grid-bias preamp stages as they limit versatility and voicing options, so I would add cathode resistors for V1A/B. This allows adding bypass caps if desired to focus the gain.

A simple fix that might tame the oscillation is to remove one or both bright caps from the output of each input stage (the 150pF across 470k feeding the respective Volume-1/2 wipers).

Since you referred to using both inputs ala plexi methods, you might want to reconfigure the mixing to just copy the plexi? On the other hand, how it is now is fairly simple although the Master Volume could be repositioned after the EQ or between the PA front-end and splitter.

Merlin's article is pretty interesting. Never had any of those problems BUT I only use self-biased concertina splitters with proper grid-stops and provide local filtering for each stage right at the circuit. Menno stressed the importance of having a high-value filter for the concertina since on signal is referenced to the supply.

Having multiple gain stages and/or cascaded stages all running from the same supply node can be problematic or not. It depends on how things are laid out. It is always preferred to have a local filter cap for each stage even if it is effectively a common supply node.This filter cap and its physical wiring are more important than resistive isolation of the supply nodes for each stage; the cap proximity is what gives "power" to decoupling by providing the return path for the signal current of the related stage - it is the decoupling.

[physics]

Thanks for the actionable pointers, I'll go start trying 'em and report back. It seems like the general idea is to reduce problematic gain so that moving the guitar closer and "closing the loop" so to speak doesn't result in oscillation taking off. Does that mean all amps couple with guitars like this, and that it's the circuit that determines whether they can handle it, not stuff like whether a transformer needs more shielding or the like?

[physics]

Ok, I ran out of time to do much more of anything with the amp before leaving town. The last two things I did after posting here was wiring up the pilot light, and rearranging a couple connections so that immediately after coming into the amp the line's hot lead goes direct to the line fuse and from there to the power switch. Previously, it went to the power switch and then the fuse, and it is my understanding that the fuse should be the absolute first thing. After that, I handed it off to my friend for play-testing and to use while I'm out of town. When I get back I'll check his notes, my notes, and start trying to fix the issues mentioned here and any others he finds.



Part of the hand-off involved renting out a rehearsal studio for a few hours to crank the amp (with hearing protection used by those who cared, myself included) and a few other amps I brought to see how they sound. Some conclusions from the max-volume testing:

• The Bogen sounds fantastic through my detuned 2x12 and has a sizeable sonic footprint. Given I didn't do anything to really shape the sound, this seems like quite the happy accident. I just rebuilt the power supply with new parts and galactic grounding, installed a low-impedance adjustable bias supply, and paralleled the mic inputs. Technically I also added some capacitance to the plate supply node, but I don't see how that would significantly impact the sound. I have some room recordings with the PZM I mentioned in another thread, after I get the ok from my friend I'll post some clips.
  
• The high-frequency oscillation I mentioned didn't seem to be an issue in an actual playing situation. At least, I didn't notice it while my friend had his guitar near the amp, playing or otherwise. Doesn't mean it's not still an issue, but it's interesting nonetheless.
  
• Blocking distortion with the bass turned up too far is an issue. Conveniently, the first fix that comes to mind is reducing the PI to power tube coupling caps, which could take care of the oscillation at the same time. Next time I have time...
  

Anyhow, that's all I have for now!

[K O'Connor]

Hi Guys

It is actually very simple why a change to the power supply filter cap values effect the tone: the power supply is the other half of the signal path.

This is exactly why Galactic Grounding is so important, and at the very least, why having filter caps close to the circuits they support is important AND why the first filter cap is ultra-important.

The main filter is the first defense against hum in the amp. It is where the pulsating DC is converted into as flat DC as possible. It is the filter that directly supports the output stage in most amps and therefore it carries the highest amplitude signal currents. This makes the capacitor impedance vital to performance.

It is no surprise that the updated amp has a big sound compared to stock.

Yes, lower-value coupling caps are the fix for many woes in tube amps.

[physics]

Hi Kevin!

To clarify, I didn't get any sound samples of the amp before digging into it. Mostly because it wasn't in a state I wanted to power it up in. That said, the amp does sound pretty different from the clip or two I've been able to find of it online, and it sounds bigger than the Peavey VK100 I was working on from what I recall.

To what you said, I do recall reading in one of the TUT's about the power supply being the other half of the signal chain, however I suppose I figured it would be more of a "feel" thing and wouldn't contribute beyond makings things stiffer or saggier. From what you're saying though it sounds like I may have underestimated the effects. I'll have to find a working amp to do some before/after tests with to get a better feel for the exact changes, since I didn't get to play this amp before overhaul. Hmm... I have a shoddily made 20W Bogen conversion I picked up a while ago that could use some loving, maybe that'll be the test bed. I may also try adding some capacitance to some nodes in my Marshall 4104 to see how that changes things. Thanks!