I have only worked with tubes when I was a kid, building powerful amateur FM radio transmitters...
That amp has almost constant distortion in all the audio spectrum, obviously due to the asymmetrical output stage.
The easiest way to decrease THD without changing anything else, is to replace the resistors on the anodes with another pair of such tubes in series, wired as constant current sources, each set at the average current of the resistor that they replace. That will significantly increase the bandwidth too and the perceived output.
As for safety, always wearing insulating gloves when building/testing/servicing, eliminates the risk of electric shock.
They are long, covering your arm, and rated at various maximum voltages up to tens of thousands of volts. Some are ridiculously overpriced, but you can find decent ones at reasonable prices -note that they should be kept in a hermetically sealed bug to maintain them, because it is latex (organic). I also plan to buy a pair soon.
I downloaded the book from that page "ESL_THEORY by FRANK VERWAAL - English_2011" which is the best I've read so far, thanks for the link.
That's interesting. Since the output for my DIY headset will be +-700V, I should make the bias also +700V, right?
For other headsets there will be a switch that will change the output voltage and bias. Also, unlike Stax and others, I make my headphones shielded and grounded with an extra cable.
First, like I said, I'll make them shielded and grounded, like my current one.
Also the overall quality of construction with safety-first, is important.
Second, a music amplifier is not a continuous energy source like the power grid. It reaches such voltages only on the peaks, momentarily, and regularly only on music climax at max volume.
Third, it depends on the amp and the connection. If the amp is a single-power supply one like my current amp at +900 volts, then with a direct connection, +450V DC will be always present on each stator, even with no sound output.
That's dangerous. You can eliminate the DC voltage if you connect the HP via capacitors with a large resistor in parallel, but that will reduce the output quality.
A +-700V amp will have zero DC voltage when there is no sound, the connection will be direct, and the danger will only be directly proportional to the volume, and the music content. For example, low music passages are in the order of a few decades of volts.
Fourth, I will implement the anti-shock safety feature I mentioned previously, so in fact, this amp and headphones, will be far safer than any other set in existence.
Last edited:
i had opened a some of my stax hp´s, internal construction - distances/materials minimizes safety risk. (unless listening in shower ..)
when building own hp, i guess there can be risk
1 rcd sensing, it could be done, but im afraid it would trip on treble transients.... high chance that a separate sensing toroids, for each stator must be built.
2 hv output caps , this makes most sense to me. eliminating also dual supplies complexity
3 maybe a hv version of dc sense out circuits used on spk amps..
on my blog i tested carbide fets, wired as hi-µ fake triode+gyrator.
basic results are nice, but must find a time to try transformer drive of this prototype
..)when building own hp, i guess there can be risk
1 rcd sensing, it could be done, but im afraid it would trip on treble transients.... high chance that a separate sensing toroids, for each stator must be built.
2 hv output caps , this makes most sense to me. eliminating also dual supplies complexity
3 maybe a hv version of dc sense out circuits used on spk amps..
on my blog i tested carbide fets, wired as hi-µ fake triode+gyrator.
basic results are nice, but must find a time to try transformer drive of this prototype
I have only worked with tubes when I was a kid, building powerful amateur FM radio transmitters...
That amp has almost constant distortion in all the audio spectrum, obviously due to the asymmetrical output stage.
I later found out that a large part of the distortion was actually due to aliasing products of the CD player I used as a sine wave source - which are emphasized by the heavy diaphragm correction. By that time Frank had already picked up the amplifier and taken it to his place, so I couldn't repeat the measurements. So all in all, I don't know the real distortion levels, I only know they must be substantially less than the values in the report.
The problem is the extra heater supply that you need then. It needs to follow the outputs to keep the cathode-to-heater voltage of the current source valves at an acceptable level. You could do that by using an extra transformer winding per current source valve and connect its centre to the corresponding output, but that means loading the output with some LC circuit consisting of interwinding capacitance and spreading inductance of the transformer.
If I had used an extra valve per side, I would have used it as part of an SRPP stage rather than as a current source. With an SRPP stage, the bias current could have been lower, which would also lead to a somewhat higher voltage out of the high-voltage cascade.
By the way, depletion MOSFETs can also be used in an SRPP-like style (but obviously without the heater problem).
The bandwidth of my amplifier was not a problem (neither the small-signal nor the power bandwidth), but the voltage was simply too low for Frank's loudspeakers. His loudspeakers used most of the signal power to heat up the resistors in his crossover network.
Where can you buy those? I used gloves meant for keeping your hands dry while you are washing up dishes - better something than nothing...
The Dutch version is even better, but I guess that won't help you much...
Disagree about "...better than nothing".
He means what we call "lineman gloves". That also includes football(?) gloves, try search "lineman gloves electrical".
An externally hosted image should be here but it was not working when we last tested it.
Lineman's Gloves - Lineman's Gloves & Protectors - Hand Protection - Safety Products - Safety - Northern Safety Co., Inc.
Hubbell Power Systems - Products for Electric Utilities, Telecommunications and Construction Industries
https://www.hantover.com/m/cgi/hnt.wsc/product_list?ccd=10303520&cat=Electrical-and-Lineman-Gloves
Note that, in the US system, the "class" implies what voltage they may be used on. Also, for power line work, they must be used with leather over-gloves to protect the rubber. There are also rubber sleeves to cover top of glove up to shoulder. These gloves MUST be tested regularly, 6-12 months depending on service.
https://us.pipglobal.com/archive/literature/Novax-FAQs.pdf
Last edited:
Oops, I didn't thought of that potential difference limitation -I never had to cascade two tubes!
That's funny, that's how my current amp works -I had arrived at the same concept after a lot of experimentation when I designed it in 2000 (with transistors), but I didn't know it had a name!
It's amazing what one can learn on such forums! Now, did a SRPP transistor version exist back then -at least on the net? I doubt it, as I hadn't seen anything like that.
I did some experiments in LTspice to refresh my memory and it's better than a current source (higher bandwidth, lower distortion), but not as good as a proper symmetrical approach -it needs to draw significantly more current to compete -if you are demanding, that is. As personally I'm aiming at ultra low distortion at double the voltage, a symmetrical approach is my only choice -and I'm leaning towards a quasi-complementary one -in steroids!
And it did work with enhancement mode mosfets too. I didn't try the depletion mode ones.
I'm considering to buy something like this (class-3 26500V)
(who knows, I might decide to make ES loudspeakers later)
Γάντια Υψηλής Τάσης (class3 -26500V) για προστασία από Ηλεκτρισμό - Pegasosafety
to be worn under a protective "overglove":
Εξωτερικά γάντια ηλεκτρολόγου Eurotechnique 2550 Overgloves - Pegasosafety
[image]The guy who made it obviously had a lot of humor!
That would be a great prank-style gift for an AV fanatic friend -if you can find it and restore it, or even make it from scratch
[/image]As far as I recall the output stage of an NE5534 is essentially an SRPP stage made with bipolar transistors.
Depletion mode is more convenient because you don't need to implement a floating voltage source, but indeed it can also work with enhancement MOSFETs (or even with bipolars).
Thanks for the advice on electrician's gloves!
You don't need a floating power source for the upper SRPP side -whatever you use. Simplicity is its advantage.
Depletion mode would be useful for the quasi-complementary topology, but for the voltages I need them >1400V, the only depletion mode mosfet available AFAIK is the IXTT2N170D2 @1700V which has an input capacitance = 3650pf (!), it is expensive ( 15.5 euros) and ...there are only 40 pieces on mouser.com!
Last edited:
You don't need a floating power supply, but you do need something to overcome the threshold when you use an enhancement MOSFET, like a decoupled voltage divider or so. That's what I meant by floating voltage source.
That depletion MOSFET indeed doesn't seem very suitable. There are depletion mode high-voltage SiC JFETs on the market, but those are also power devices with large capacitances.
That depletion MOSFET indeed doesn't seem very suitable. There are depletion mode high-voltage SiC JFETs on the market, but those are also power devices with large capacitances.
You only need a pull-up resistor from the gate to the power supply (V+). And you only need a zener and a resistor to complete the circuit that will monitor the lower semiconductor's current and will make the opposite push-pull action.
Although I haven't used depletion mosfets yet, I don't see any particular advantage in high-voltage amps, because they too have a threshold to overcome, in order to turn them off, and after pulling the voltage down it should also comeback up -which is all the trouble.
Whatever you can do with one, you can do with the other with the same pull-down action from the current driver (from zero to negative to make a depletion one off, from positive to zero to make an enhancement one off) although there is a difference on how and when that drive current is consumed in each one. They might simplify a bit some specific topologies and applications, but nothing serious or revolutionary.
Whether you'll use a floating supply or not, depends on a combination of other factors, like the speed you require the device to respond vs its input capacitance, the max power dissipation of the device vs the drive current and the voltage difference, and in short whether your demands are covered or not without one.
Thanks, they seem great, but they don't satisfy my voltage requirements above 1400V -the "safe operating area" graph for some reason shows DC operation bellow 1000V and above 100ma -that's a very linear-unfriendly graph, obviously because they are meant for switching apps only -as advertised. Plus, the company has closed several years ago. I would rather not invest any effort in "ghost" companies and parts.
Last edited:
Interesting; except it appears that SemiSouth closed shop 5 years ago, and I am not finding any stock for sale.
Semiconductor Today- 24 October 2012 - SemiSouth to be closed; 90 staff to be laid off
It's more fun than I expected...
So far I've read contradictory articles about components and technologies -both official and from individuals (frustration), I've seen amp schematics that would blow up if a continuous 20Khz waveform was applied, (entertainment) and have seen specific prospect components from big distributors to disappear -all at once! (mystery)
Worst part is the negative temperature coefficient of mosfets, that is, for the same gate-source voltage, the current will increase multiple times with temperature. Which means that hotspots can quickly be created on the die (which is consisted of many cells), that absorb more and more current due to the negative coefficient and the part can die cold! (terror)
That is, when we use them for the exact purpose we are talking here: linear applications! (disappointment)
Otherwise, I had some preliminary virtual success (with LTspice - not sure how far from reality) reducing THD down to 0.025% at 20KHZ -1400Vpp output and 250pf load (fun) without even starting to use a quasi-complementary topology along with sophisticated techniques.
Still, I have to reduce that number 25 times as my new goal is now 0.001% (instead of 0.01%), solving all problems of course (challenge)
Not an easy task though, as that means distortion in the order of 1/100,000 -a really clean waveform down to microscopic level. That's a long distance from the current 1/4000...
So, is that (THD+noise) goal possible (in a reasonable R&D time) at 1400Vpp, 0-20KHZ? (suspense)
So far I've read contradictory articles about components and technologies -both official and from individuals (frustration), I've seen amp schematics that would blow up if a continuous 20Khz waveform was applied, (entertainment) and have seen specific prospect components from big distributors to disappear -all at once! (mystery)
Worst part is the negative temperature coefficient of mosfets, that is, for the same gate-source voltage, the current will increase multiple times with temperature. Which means that hotspots can quickly be created on the die (which is consisted of many cells), that absorb more and more current due to the negative coefficient and the part can die cold! (terror)
That is, when we use them for the exact purpose we are talking here: linear applications! (disappointment)
Otherwise, I had some preliminary virtual success (with LTspice - not sure how far from reality) reducing THD down to 0.025% at 20KHZ -1400Vpp output and 250pf load (fun) without even starting to use a quasi-complementary topology along with sophisticated techniques.
Still, I have to reduce that number 25 times as my new goal is now 0.001% (instead of 0.01%), solving all problems of course (challenge)
Not an easy task though, as that means distortion in the order of 1/100,000 -a really clean waveform down to microscopic level. That's a long distance from the current 1/4000...
So, is that (THD+noise) goal possible (in a reasonable R&D time) at 1400Vpp, 0-20KHZ? (suspense)
Do your MOSFETs suffer from Spirito instability (which is basically the same as thermal second breakdown: self destruction by local heating causing increased local current causing even worse local heating and so on), or is it only the overall current at a given VGS that increases too much? In the latter case, you might be able to solve it by changing your circuit design, by adding source degeneration for example (or by any other technique that senses and controls the drain current). With Spirito instability you can only look for a more suitable MOSFET type with a better SOAR graph.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.