high side H bridge signal chopper

Thread Starter

Colin Leamy

Joined Dec 28, 2017
6
There's a project I want to build that I saw online a few years ago, unfortunately the poster never updated his progress, I was hoping to get some insight here as to how I might go about constructing this project.
The project in question being explained here http://tubelab.com/articles/ideas/solid-state-opt/

As far as I understand it splices up the signal amplitude at a constant frequency outside the audio band and then unsplices it at the other end of the transformer. What I don't get so far is how the displayed circuit on the left is supposed to function because to me it looks like it will effectively just act as a high frequency source follower on the load. I don't see how you can make transistors act as pure switches in this way.
 

Janis59

Joined Aug 21, 2017
765
Who is using the vacuum tubes in nowadays? Only two way of guys - me because at 100+ MHz today is not yet invented the Mosfets with power of 200+++ Watts, and even then I try always to find how to shift away from tubes. And audi-maniacs those kinda speaking on `lamp sound`/`tube sound`. Yes, it is true that phase noise in vacuum lamps are far better than any even best semiconductor device, BUT if You use ANY semiconductor in such circuit, You have lost all the tube`s benefits, kept all tube`s weaknesses, and gained all semiconductor`s weaknesses. Why to act so unwise?
If Your aim is to get the megawatts output, use the mosfets brutally parallel (even hundred of them or more), if the aim is to get the maniacally great sound quality (sorry, in my childhood the elephant was stepped over my ears so I am not able to see ANY big difference between fish-can quality sound and Hi-Fi ++ extra super duper), then let use those radio-lamps (tubes) as much You want, only obtain the carriage on wheels beforehand and not mix them with semiconductors - nowhere in sound track, other-how the system`s quality is governed by the weakest system`s element..
 

Thread Starter

Colin Leamy

Joined Dec 28, 2017
6
Did you seriously just turn this into a tube vs solid state thread? I could give you an intelligent response but I won't.
Can we please stay on topic.
 

Janis59

Joined Aug 21, 2017
765
Did you seriously just turn this into a tube vs solid state thread? Can we please stay on topic.
I was giving You hint that You try to get together a fire and water. If still will, go ahead. Yet You haven`t stated the problem.... WHAT is that You don`t see how to realize?
If recommendations on what semiconductors? Probably the IXFH42N60 is the brilliant option with some 20++ Amps and 600 Volts what by datasheet ought be applicable up to teen of MHz but at praxis stands nearer to 100 MHz. At least 40-70 MHz it still works rather good.
If question was about how to steer the gates, then there is no other answers as - or optron, or ferrite core gate transformer.
If there was question on output filter - there I have no a large experience. If question was about topology - be wildly aware about EMI (means groundwork appropriateness), that is always problem when commuting MHz in hundreds of Volts.
 

Thread Starter

Colin Leamy

Joined Dec 28, 2017
6
My issue mostly lies with how it is supposed to work.
Actually the more I think about it the more I don't understand the topology at all.
It looks like an H bridge to me, but the mosfets will just project their vgs voltage and signal onto the triode, like a source follower, almost completely negating any signal from the triode.
Also the upper mosfets are p channel wired in reverse it seems.
I have no idea how that circuit is supposed to chop up the output signal of the triode, especially acting as a load to the triode.
 

Janis59

Joined Aug 21, 2017
765
The only meaning of circ is to convert Your kHz signal into MHz signal what demands some 100-fold smaller core for transformers.
However, there always stands the sacramental question where to get the cores qualified for MHz and kW for size You need (what is far far biger than E55-E65). The producers at US/Canada - damn expensive, at russia - what is most strange, roughly the same price than US but much less trustable. Or India, what being the Mecca for appatites (Cr and Ni) has the most good start positions. There I found the certain firm, producing the 1x1x4 inch i-cores, what is VERY easy composable to ANY size and/or geometry of core (with epoxy resin). After payment to their EU represaentative where they footballed me, I was waiting near the half of year to receive it (be warned!), but 0,39 Tesla at 3 MHz is impressive indeed, so I am now very satisfied with about 1 USD per piece (what in case of USA would be 15 and russia/St.Petersburg 12 USD). My case the need was 100 kW, so I bought a full palette.
 

Thread Starter

Colin Leamy

Joined Dec 28, 2017
6
Forget the transformers, it's not currently relevant. The tubelab guy pulled a random one out of a PC power supply and used it at 70khz and it was sufficient. It's not my concern at the moment.

The only meaning of circ is to convert Your kHz signal into MHz signal what demands some 100-fold smaller core for transformers.
Yeah I get that part. It transfers the signal amplitude across a constant frequency.
I don't understand how, for the reasons I listed above.
I simply doesn't work.
 

Janis59

Joined Aug 21, 2017
765
Oh, again I forgot to answer Your main question,- it shifts up and down the polarity sequentially. So, left upper and right lower are opened together, and upper right with lower left together when first pair is closed. Don`t forget to install the proper time-shift between the gate clk phases, to avoid the through-shoots.
 

Janis59

Joined Aug 21, 2017
765
1) 70 kHz means that Your upper audio frequency will be not much higher than some 5 kHz. For tram`s audio its OK. For Hi-Fi - the most precise name is `diversion` (in finely military meaning of word).
2) How it works You may read on keywords of E-class-amplifier, H-class-amplifier
 

Thread Starter

Colin Leamy

Joined Dec 28, 2017
6
Yeah I get the theory of how it is supposed to work. But the issues I stated above still stand.
Here is the spice file I made of it. It completely doesn't work at all.
 

Attachments

Janis59

Joined Aug 21, 2017
765
I tried to open file in my LT v4 but it contains many objects missing. Probably that is gate driver (amplifier). Then 1) any trafo demands VERY hard demands about symmetricity of chop, thus the frequency divider to two after 555 is mandatory (4013). But that is the smallest thing there. I see the main reason is that You brutally connected one gate with +4V level with other with level of probably 600V. That is what demolishes the circuit in model, and in the nature unavoidable must make a lot of smoke. Just apply the optron or gate trafo for galvanic barrier and get be blessed.
 

Thread Starter

Colin Leamy

Joined Dec 28, 2017
6
Oh yeah sorry, forgot to bias the mosfets in that one.
Here is one with mosfet amplifier instead of triode.
I used ideal mosfet drivers @ 70khz
The mosfets completely corrupt the output of the amplifier.
 

Attachments

Top