Huh? You mean a spark gap that is not shielded against atmosphere? Although there are many types of "open" spark gap devices still sold they would not be reliable in this application.
Gas Discharge Tubes have been around for 100+ years. The GDT is designed to fire at a pre-determined voltage and survive multiple firings. There are billions sold each year and they are dirt cheap. The cheap ones work from 75-2000 VAC and come in various power ratings and with 2 or 3 terminals for differential lines. The cheapest spark gaps are just traces on a PCB. Again not as reliable as a GDT.
In the old days GDTs with a radioactive gas fill were preferred but that is another subject.
Huh, shield air against atmosphere? In what kind of atmosphere do you run your amp?
And what make you think a open air spark gap is unreliable?
On a PCB they may be of limited use and unreliable for sure but properly done they are as reliable as it can get if you know how to do it.
Last edited:
The break-over voltage can vary over a wide range due to mechanical tolerance, humidity, contamination, altitude... Not desirable for electronics. OK for large equipment where +/-50% error won't fry anything. With a GDT being so cheap why bother. Certainly no low cost GDT can match a spark gap for power rating.
I have a surge generator in my shop (5KV@500A). It will blow semiconductors and even wire-wound 50W resistors into pieces.
Last edited:
Contamination and mechanical tolerance can easily be handled by a proper design and build up.
When it comes to humidity and altitutude the open air gap has an ADVANTAGE because the transformer will most probable WORK UNDER THE SAME CONDITIONS .
Assume for instance that for some reason the airs breakdown voltage goes down , so will the open air transformers capability to withstand over voltage.
The open air gap break down voltage will go down also and therefore still protect, whilst the fixed voltage GDT wont change a yota and protection is more uncertain.
Or would you prefer a fixed voltage over one that is essentially self adjusting to the transformers capability?
Last edited:
Fellows (and lasses)
The “trick” is to keep the dissipated power in a pair of series-connected 600 (or 1000) PIV diodes well moderated during the spikes. The first obvious plan is the pair of diodes (reversed) in series with an inductor, in series with an appropriate resistor. The choke squashes the fast risetime flashover, the resistor perhaps takes the lion's share of the spike(s) dissipation.
It is however not a willy-nilly design spec to optimize. The actual peak-inverse-voltage breakdown needs to be known, and the spec for the chokes has to be both economical and yet high-enough in henries to moderate the spikes. While still passing most of the spike energy thru the resistor-and-breakdown-diode. Balancing trick.
One might argue to increase device dissipation (diode-side) by using a chain of Zeners. Maybe use 70 volt units in series. Good trick. Kind of spendy though if too many are utilized. But hugely predictable.
Just saying,
GoatGuy ✓
The “trick” is to keep the dissipated power in a pair of series-connected 600 (or 1000) PIV diodes well moderated during the spikes. The first obvious plan is the pair of diodes (reversed) in series with an inductor, in series with an appropriate resistor. The choke squashes the fast risetime flashover, the resistor perhaps takes the lion's share of the spike(s) dissipation.
It is however not a willy-nilly design spec to optimize. The actual peak-inverse-voltage breakdown needs to be known, and the spec for the chokes has to be both economical and yet high-enough in henries to moderate the spikes. While still passing most of the spike energy thru the resistor-and-breakdown-diode. Balancing trick.
One might argue to increase device dissipation (diode-side) by using a chain of Zeners. Maybe use 70 volt units in series. Good trick. Kind of spendy though if too many are utilized. But hugely predictable.
Just saying,
GoatGuy ✓
http://www.teledynereynolds.com/resource-library/Documents/Spark Gaps.pdf
Digi Key caries three manufacturers. Lots of devices available.
Digi Key caries three manufacturers. Lots of devices available.
Last edited:
Well … yes … then there are the gas-filled units. Basically bomb-proof. If you think about it (and just so happen to have a lot of them) you could easily deploy a series chain of inert-gas voltage regulator tubes, as well. They can handle outrageous abuse of short-lived spikes, if need be.
GDT means Gas Discharge Tube. The key to their power handling capacity is once triggered, the voltage across them is very low. Of course this is the same for a open air spark gap or similar to a Thyristor. A voltage regulator tube cannot handle the anywhere near the same power unless it was massive. A small thyristor cannot handle the current of a GDT which is >5,000 Amps. The downside to a GDT is the 1uS turn on time. There are commercial GDTs with radioactive gas fills and HV bias voltages that turn on much faster. Very pricey.
And how on earth would even a choke damped by a resistor protect the transformer from its voltage spike when exactly the opposite of it, a resistor damped capacitor would be needed? A properly sized cap and resistor makes sense, but a choke would only decrease the current rise time to the diode and during that time the transformer voltage would only rise instead of getting limited.
Please explain how this "spashing" is supposed to work.
Last edited:
Hmmm… I think we're veering off-thread.
But I thought of another problem too (still sort of "on thread"), and that is that once triggered, gas-discharge devices tend to stay conducting until both a minimum holding current and voltage is not met. I think in the service of the intended device, to protect … a vacuum tube amplifier output … well, getting the darn thing to turn off might be a challenge.
Just saying,
GoatGuy ✓
And — pardon for borrowing, Mr. Wavebourn — it looks like Wavebourn already outed a good answer that works, too. Cheap, effective, purpose built.
Just saying,
GoatGuy ✓
In the railway industry, spark gaps with carbon disks are used to protect lines. Some are not sealed. Railroad Lightning Arrestors
There are numerous other spark gaps and GDTs used for industrial applications. Unless you work in these industries you would never see them. In you are a HV hobbyist, the rotary spark gap is quite common.
This topic crops up every now and again. Technology advancement has helped over time.
I have a vintage PA amp with nicely formed copper wire to provide a P-P airgap, as was advocated by GEC in their 1957 audio amp booklet. I have a slightly latter amp with a 6AL3 damper diode from each plate to ground - a valve version of the 3x 1N4007 in series flyback diode scheme in common use in the 1960's. Vintage Dynacord amps used a GDT from P-P, and Traynor amps used Thyrectors.
With respect to applying a GDT, its arc initiation voltage is quite dependent on frequency and waveshape, and is better applied P-P due to its latching character.
An amp's output stage is effectively a 'low energy' environment, and significantly different from applying protective devices for mains side circuits - so a good awareness of why and how that matters is important, especially if waxing lyrical about what certain protective devices can and can't do.
Imho, the cheapest and simplest form of protection is a modern small MOV across each plate-to-B+ winding. MOV's are plentiful in voltage ratings aligning with mains ACV levels, and can be connected in series to obtain a Vdc (1mA) rating that is sufficiently above any normal plate to B+ voltage excursions, but with a clamping capability below circa 2kVdc that would by at the low end of transformer insulation capabilities. They are benign to circuit operation normally, and if they do start to influence operation then it is initially very soft.
Sometimes protection is not needed, because other circuitry already loads a plate node (eg. amps that use plate feedback, or RC frequency shaping, or limiter circuitry).
I have a vintage PA amp with nicely formed copper wire to provide a P-P airgap, as was advocated by GEC in their 1957 audio amp booklet. I have a slightly latter amp with a 6AL3 damper diode from each plate to ground - a valve version of the 3x 1N4007 in series flyback diode scheme in common use in the 1960's. Vintage Dynacord amps used a GDT from P-P, and Traynor amps used Thyrectors.
With respect to applying a GDT, its arc initiation voltage is quite dependent on frequency and waveshape, and is better applied P-P due to its latching character.
An amp's output stage is effectively a 'low energy' environment, and significantly different from applying protective devices for mains side circuits - so a good awareness of why and how that matters is important, especially if waxing lyrical about what certain protective devices can and can't do.
Imho, the cheapest and simplest form of protection is a modern small MOV across each plate-to-B+ winding. MOV's are plentiful in voltage ratings aligning with mains ACV levels, and can be connected in series to obtain a Vdc (1mA) rating that is sufficiently above any normal plate to B+ voltage excursions, but with a clamping capability below circa 2kVdc that would by at the low end of transformer insulation capabilities. They are benign to circuit operation normally, and if they do start to influence operation then it is initially very soft.
Sometimes protection is not needed, because other circuitry already loads a plate node (eg. amps that use plate feedback, or RC frequency shaping, or limiter circuitry).
Last edited:
It would never be prudent to use a GDT, spark gap or Thyistor across a power feed unless the PS had a "hiccup" protection mode that allowed current to drop below the holding current.
A average size Varistor cannot handle anywhere near the power that a GDT can. Also a GDT won't explode and/or catch fire like a Varistor will. They are nasty beasts when overloaded.
In this application the use was plate-plate where a GDT would be OK and I'm sure a Varistor would be too.....
I forgot to mention the ubiquitous spark gap capacitor. Widely used in the primary side of switching power supplies and was very common in CRT bias feeds.
That is an interesting point. It seems that if the secondary circuit went open and the primary started to rise (with DC plate to cathode FB for example) the rising voltage would be shunted by Rk and Rf thus limiting the voltage rise. That is a good reason to DC couple the FB. One would want to factor that posible extra current and voltage into the ratings for those resistors.
With traditional interstage and traditional output transformers:
Global Feedback from a transformer secondary, or any other feedback from a transformer secondary is not exactly DC coupled.
But that method works very well for thousands of amplifier designs.
Sometimes you have to find other solutions of arc-over protection.
If you want DC coupled feedback from the output stage that has a transformer, then take the feedback from the plate or primary winding.
Extra Credit:
What transformer is DC coupled from input to output? . . .
A Common Mode Choke, DC is coupled through the individual windings
(and it is AC coupled from one winding to the other winding).
Global Feedback from a transformer secondary, or any other feedback from a transformer secondary is not exactly DC coupled.
But that method works very well for thousands of amplifier designs.
Sometimes you have to find other solutions of arc-over protection.
If you want DC coupled feedback from the output stage that has a transformer, then take the feedback from the plate or primary winding.
Extra Credit:
What transformer is DC coupled from input to output? . . .
A Common Mode Choke, DC is coupled through the individual windings
(and it is AC coupled from one winding to the other winding).
Last edited: