I was looking at an old article/schematic and noticed it had fuses connected to the output cathodes. I don't think I've seen that generally done - is it because fuses are too non-linear compared to a low value resistor? I like the idea of an easily replaced part if a tube goes south.
Attachments
No real concern with using a fuse as far as linearity. A 1/8 amp fuse may be problematic as far as getting with a reasonable voltage rating. You would need a fuse rated for 500 or so volts. A regular 250 volt fuse may internally arc and make a mess if it blows and when fed closer to 500 volts. Also there could be false failures with such a tiny fuse current.
Better to fuse the B+ at the transformer though you need a high voltage fuse.
We usually just supply a 10 ohm cathode resistor for measuring bias and hope it dies if there is a bit much over current due to either a bad tube or bias voltage.
Better to fuse the B+ at the transformer though you need a high voltage fuse.
We usually just supply a 10 ohm cathode resistor for measuring bias and hope it dies if there is a bit much over current due to either a bad tube or bias voltage.
That amplifier has a serious liability. In the interest of achieving better low-frequency stability, the driver plates are effectively direct-coupled to the output tube grids. A percentage of the drivers' positive plate voltage appears on the grids of the 6550s. It is counterbalanced by a larger-than-usual negative bias supply that compensates for the positive voltage and fixes the grids at the proper negative voltage value. However, if the bias supply fails for any reason, the 6550s will see positive voltage on their grids and will begin running red hot. The fuses are there to (hopefully) prevent the tubes from melting down, which they will in about 10 seconds.
I've built that amp, and not only does it not sound nearly as nice as a standard, cathode-biased Williamson, but it always made me nervous as an accident waiting to happen.
I've built that amp, and not only does it not sound nearly as nice as a standard, cathode-biased Williamson, but it always made me nervous as an accident waiting to happen.
I had a normal 250v glass fuse in cathode of a 2500v plate triode
when it blow the glass exploded, really dangerous
when it blow the glass exploded, really dangerous
It says a '1/8 amp cathode fuse' in the schematic - did there used to be a special type of fuse for this purpose?
Small film resistors is linear until the moment they blow, and they blow instantaneously which a fuse does not.
( see "flame-proof" resistors that will not gutter the innards when flaming)
( see "flame-proof" resistors that will not gutter the innards when flaming)
I'm really curious about your experience with the amp - the design is from an old Herbert Keroes article and seems to be the basis for some of the Dynaco designs.
I took a brief look at fuse rated resistor voltage ratings (Tyco FRN25) and they seem to be limited to <500V and it doesn't solve the easily replaceable part for an end-user? I'm trying to figure out if I can build an amp that I can give to my kids that won't require opening up and desolder/solder in the event of a tube going bad.
The cathode fuse won't be exposed to 500V, and this isn't a 2500V triode related amp.
The cathode voltage would rise, but that pushes the cathode into deep cut-off - also note that Vhk may start to leak above 200V and there appears to be about +45V heater elevation - so my guess would be the cathode likely wouldn't rise to stress Vhk, and so the tube may well be saved (if it wasn't the culprit). If there was still concern, then paralleling each fuse with say an 80V 5W zener would be reasonable imho.
The cathode fuse is unlikely to experience turn-on stress like a normal mains fuse, or secondary CT fuse, so that mitigates one form of false failure. 125mA fuses are a common rating for 5x20.
A fuse in the secondary side CT can adequately be a 250VAC rated fuse - that advice was once part of Littlefuse design advice, but it seems they removed the relevant comment a few years ago, perhaps from legal advise. Fusing the B+ is ugly wrt to dc voltage and lots of inductive parts (choke and OPT primary).
The cathode voltage would rise, but that pushes the cathode into deep cut-off - also note that Vhk may start to leak above 200V and there appears to be about +45V heater elevation - so my guess would be the cathode likely wouldn't rise to stress Vhk, and so the tube may well be saved (if it wasn't the culprit). If there was still concern, then paralleling each fuse with say an 80V 5W zener would be reasonable imho.
The cathode fuse is unlikely to experience turn-on stress like a normal mains fuse, or secondary CT fuse, so that mitigates one form of false failure. 125mA fuses are a common rating for 5x20.
A fuse in the secondary side CT can adequately be a 250VAC rated fuse - that advice was once part of Littlefuse design advice, but it seems they removed the relevant comment a few years ago, perhaps from legal advise. Fusing the B+ is ugly wrt to dc voltage and lots of inductive parts (choke and OPT primary).
Ordinarily the cathode may see about 50 volts and a typical small fuse will work OK. If the fuse is intended to protect the tube and possibly the OPT from a bias failure this scenario will work. Unfortunately, this is not the only common failure mechanism.
Most tubes produced during the "good old days" had a good vacuum and highly purified internal components. The usual failure mechanism was loss of cathode emission from age. Some of the new production stuff has a less than optimum vacuum and internal purity. This may not be too much of an issue with small signal tubes. Output tubes that run hot are a different story. Heat will cause the impurities to slowly "outgas" into the vacuum creating ions which can cause grid current that slightly offsets the grid bias. This causes the tube current to increase which adds more heat. More heat, more gas. Some tubes will just slowly die young from this, but some will go into a runaway condition, particularly in a class A amp. Once a tube goes into runaway the current and heat will continue to increase until something blows, or the tube internally arcs over. A tube arc can put the plate voltage on every element in the tube.
A fuse in the cathode may stop the runaway before it can reach flashover (arc). In this case a small fuse will also work. Once the tube's grid loses control of the plate current there is a very short time window before something blows. This has been the biggest failure mechanism with Tubelab SSE amps that run a tube hard in class A. The usual failure is a open cathode resistor which puts a lot of voltage on the bypass cap making for a dead cap and sometimes a leak or venting situation. In most cases the runaway takes a long time (hours) to get to the critical stage, and a fuse in the cathode should stop the runaway before other damage occurs. Beware that a typical glass fuse can explode violently when fed high voltage DC. If you fuse the cathode, put the fuse in a place where fragments cannot reach the user. Granted this is a low probability of occurrence, but one that can have unwanted consequences.
NEVER put a fuse in a DC line that is not rated for the full DC voltage that it must eat when it opens. A DC arc does not have an opportunity to quench when the voltage crosses the zero axis like AC does. A 1/4 inch glass (3AG) fuse will explode like a big firecracker when it blows on DC. A 250 volt AC fuse in the B+ to the OPT is the worst case. Been there, blown that! Scared the crap out of me. Never do that again!
Most tubes produced during the "good old days" had a good vacuum and highly purified internal components. The usual failure mechanism was loss of cathode emission from age. Some of the new production stuff has a less than optimum vacuum and internal purity. This may not be too much of an issue with small signal tubes. Output tubes that run hot are a different story. Heat will cause the impurities to slowly "outgas" into the vacuum creating ions which can cause grid current that slightly offsets the grid bias. This causes the tube current to increase which adds more heat. More heat, more gas. Some tubes will just slowly die young from this, but some will go into a runaway condition, particularly in a class A amp. Once a tube goes into runaway the current and heat will continue to increase until something blows, or the tube internally arcs over. A tube arc can put the plate voltage on every element in the tube.
A fuse in the cathode may stop the runaway before it can reach flashover (arc). In this case a small fuse will also work. Once the tube's grid loses control of the plate current there is a very short time window before something blows. This has been the biggest failure mechanism with Tubelab SSE amps that run a tube hard in class A. The usual failure is a open cathode resistor which puts a lot of voltage on the bypass cap making for a dead cap and sometimes a leak or venting situation. In most cases the runaway takes a long time (hours) to get to the critical stage, and a fuse in the cathode should stop the runaway before other damage occurs. Beware that a typical glass fuse can explode violently when fed high voltage DC. If you fuse the cathode, put the fuse in a place where fragments cannot reach the user. Granted this is a low probability of occurrence, but one that can have unwanted consequences.
NEVER put a fuse in a DC line that is not rated for the full DC voltage that it must eat when it opens. A DC arc does not have an opportunity to quench when the voltage crosses the zero axis like AC does. A 1/4 inch glass (3AG) fuse will explode like a big firecracker when it blows on DC. A 250 volt AC fuse in the B+ to the OPT is the worst case. Been there, blown that! Scared the crap out of me. Never do that again!
I used small 10R resistors instead of the fuses. I've been fooling with Williamson designs for 5 years now, and I had some Acro TO-330s in the shed and wanted to see what more power would sound like. I didn't care for it and went back to the usual 25wpc cathode-bias designs, which is really enough power for me and much easier to manage.
More the point, though--IMO this is not an amp to build for your heirs. Aside from the dicey nature of the grid-bias circuit, the chances of finding TO-330s (or Dynaco A-430's, which were the other option for this amp) are slim to none, in which case you're stuck with a substitute that will require re-designing the feedback compensation, which is a lot of work and won't guarantee stability. Even built as specified, the amp has excessive high-frequency bandwidth and, despite the author's claims, there's a significant amount of ringing going on. This is pretty much the last gasp of the Williamson, and better circuits came along that offered more stability and power without jumping through hoops.
Don't get me wrong, I love my Williamson amps but past 25 wpc there are better ways to build a mousetrap. If I were going to build a trouble-free tube amp for my kid, I'd look at an Eico HF-87 clone, which is cathode-bias, about as reliable as they come, and offers a very pretty 35wpc. There's a design on another site that uses Hammond OPTs and has been tested and stabilized. I'm sure there are other options as well that people will suggest.
In the end, I'm not sure there's ANY tube amp design that will guarantee no-brainer fixes for tube failures, but this wouldn't be a prime candidate for me.
Did i misunderstand the above ?
Dynaco A430 is in new production, dynakitparts.com is one distributor. Also Any Dynaco MkIII is a possible donor
for a A430 transformer.
Dynaco A430 is in new production, dynakitparts.com is one distributor. Also Any Dynaco MkIII is a possible donor
for a A430 transformer.
I haven't seen an A-431 in stock at Dynakit Parts for ages. Triode USA claims to have them--for $800 a pair! And I wouldn't tear down a working Mark III to build this amp.
All the input has been very helpful - I've seen cathode fuses in guitar amps (i.e. JCM900) but I hadn't seen it elsewhere. Borrowing from the JCM900, I think I'd put a resistor and led across the fuse to keep the cathode from floating and as an indicator.
I have a pair of modern A431 copies I've been tinkering with
I have a pair of modern A431 copies I've been tinkering with
In that case, I suppose you could give it a shot. ;-) For the A-431, you might want to look at the schematic in the Dynaco catalog:
http://knob.planet.ee/kirjandus/books/Dynaco_Transformers.pdf
It uses EL34s, two chokes in the power supply and a different setup for the bias supply. The Keroes circuit allows you to balance the two output tubes but not adjust the bias itself. The Hafler circuit allows you to adjust the bias but not balance the two tubes. The Hafler circuit also has a different phase lead cap across the feedback resistor, which is the one to use for the A-431. NEITHER circuit has a low-pass filter at the input to limit the HF response, but the A-431 has a smoother HF response than the TO-330, so is easier to work with. In both cases you'll need to improvise a way to measure the cathode current, as you've noted.
I've read differing opinions on sourcing the bias supply from the power tranny secondaries or center tap, as opposed to a separate bias tranny. A separate tranny seems to be more trustworthy, in the case of an HV winding failure or short. Something like that... ;-)
Whichever one you build, be sure you observe the resistor values of R17 through R23. This is absolutely critical, since together they form a voltage divider that determines the amount of positive voltage applied to the output tube grids from the driver stage plates. Whatever you do, don't mix and match them! Choose one design or the other. Sorry to harp on this, but a friend of mine, an otherwise experienced DIYer, did this and came close to a catastrophic meltdown when he first fired up the amp.
http://knob.planet.ee/kirjandus/books/Dynaco_Transformers.pdf
It uses EL34s, two chokes in the power supply and a different setup for the bias supply. The Keroes circuit allows you to balance the two output tubes but not adjust the bias itself. The Hafler circuit allows you to adjust the bias but not balance the two tubes. The Hafler circuit also has a different phase lead cap across the feedback resistor, which is the one to use for the A-431. NEITHER circuit has a low-pass filter at the input to limit the HF response, but the A-431 has a smoother HF response than the TO-330, so is easier to work with. In both cases you'll need to improvise a way to measure the cathode current, as you've noted.
I've read differing opinions on sourcing the bias supply from the power tranny secondaries or center tap, as opposed to a separate bias tranny. A separate tranny seems to be more trustworthy, in the case of an HV winding failure or short. Something like that... ;-)
Whichever one you build, be sure you observe the resistor values of R17 through R23. This is absolutely critical, since together they form a voltage divider that determines the amount of positive voltage applied to the output tube grids from the driver stage plates. Whatever you do, don't mix and match them! Choose one design or the other. Sorry to harp on this, but a friend of mine, an otherwise experienced DIYer, did this and came close to a catastrophic meltdown when he first fired up the amp.
The Kereos 1955 schematic in post #1 shows one way that was used in the 1950's to improve low frequency stability margin - dc coupling the driver stage to the output stage, which as indicated is a somewhat dangerous approach to take. That issue is nowadays dealt with by reverting to the standard capacitor coupling between those stages, and even reducing the value of C down to say 0.1uF, but substantially increasing the value of coupling cap used between PI and driver stage to circa 1 to 2uF (not a part that was practical to deploy back then).
I use fuses in my el34 se amp fixed bias. Fusing works has saved my opt transformer and tube a couple of times.i have 2200 ohm across fuse so when it blows cathode is still ref to ground. By the way no sonic penalty for using fuse.have done ab test with and without. No change sonically. Im using generic fuse.non of that audiofile fuse bs.