clarifications
The only way to see what current that would be is to look at the transfer charts for that particular device... or just build the circuit and find out (use large heatsink).
The key point is that in a follower situation the voltage between gate and source track each other... so in actuallity this voltage difference probably never approaches 20v... (only maybe in VERY heavy clipping or transients). Nelson Pass or Hugh can explain better than I.
Other than that.......
The only way to see what current that would be is to look at the transfer charts for that particular device... or just build the circuit and find out (use large heatsink).
The key point is that in a follower situation the voltage between gate and source track each other... so in actuallity this voltage difference probably never approaches 20v... (only maybe in VERY heavy clipping or transients). Nelson Pass or Hugh can explain better than I.
Other than that.......
Pan said:I thought that a FET lookig into a inductor would be like a short at DC current. Close to it. Less than 100 milliohms.
In the SOZ for example. I understand that no matter how high you bias the FET´s the power resistors ultimately limits the current, and the FET is "safe".
I thought there was a need for something to "control" the current more than the inductor and it´s DC resistance and the bias voltage at the gate. The inductor dc voltage drop is maintained at 325 mV by a dc feedback loop. This corresponds to te 3.5 amps.
I´m no EE appearantlySo what? Neither am I!
A question: If a FET is connected to a "single ended" power-supply (a battery for instance), and a voltage divider put on half the PS Voltage to the gate of the FET, what happens? In the case here the current starts going up at about 10 amps per volt according to the dc resistance of the choke.
Is this a "CCS" and in such case, how do I calculate the current? It is a CCS but the gate threshold voltage goes down with rising temperature causing the fet to turn on harder. You have to compensate for that. I didn't calculate the current as such, just set up a cct on the bench and gave it enough gate voltage so I got the current I wanted.
Guess I should get some good books and refresh my memory on how electronics works
/Peter
Re: Re: DC offset
You gave me a good idea. I was planning to design a down-firing subwoofer. That thing, i.e. a non vertical driver suffers from gravity, and is g-biased, isn't it? if I correct this DC biasing, I will have the cone centered again, right?
Yeah, I now, is a stupid thing... I'm working and needed a relax.
Circlotron said:
You gave me a good idea. I was planning to design a down-firing subwoofer. That thing, i.e. a non vertical driver suffers from gravity, and is g-biased, isn't it? if I correct this DC biasing, I will have the cone centered again, right?
Yeah, I now, is a stupid thing... I'm working and needed a relax.
Let's assume the gate is sitting at +5v and the source is at almost zero V. If the gate then swings from +5v to +25v, a change of 20V then the source will *follow* it, going from 0V to +20v. Conversely, if the gate is driven 20V downward from +5V to -15V then the choke sees it's big chance to pull the source down by the same 20V (and the loudspeaker at the same time), from 0V to -20v.Pan said:
Ohhh.. just realised exactly what you mean....
when the gate swings downward the choke pulls the source downward by an equal amount so the gate never actually catches up with the source and clips.Circlotron
I think I´ll have to start eating Q10, Ginko Biloba or something to enhance the function of my brain
If I read your post right then if the source swing from 0V to -10V then the mosfet is NOT turned off as the voltage pass the critical turn on/off voltage of 2-4V?
Is it so, that if the mosfet is turned on with a 5V gate bias, and the drain/source idle current once is flowing, then the mosfet really is ON, and will not turn off just because the 5V gate bias is lowered?
I´ll send you a box of chocolate once you have helped me understand the basics of mosfet operation
/Peter
I think I´ll have to start eating Q10, Ginko Biloba or something to enhance the function of my brain
If I read your post right then if the source swing from 0V to -10V then the mosfet is NOT turned off as the voltage pass the critical turn on/off voltage of 2-4V?
Is it so, that if the mosfet is turned on with a 5V gate bias, and the drain/source idle current once is flowing, then the mosfet really is ON, and will not turn off just because the 5V gate bias is lowered?
I´ll send you a box of chocolate once you have helped me understand the basics of mosfet operation
/Peter
mmm... chocolate...
What really matters here is the gate-to-source voltage. Let's assume that at rest the g-s voltage is exactly 5v. Then, if the source swings from 0V to -10V, the gate would need to swing from +5v to -5v to let this happen. They have both gone down by 10v, so the gate is still 5v above the source. In reality it would be something like 4.9v above the source because the g-s voltage has to go down a little to reduce the fet conduction.
The g-s voltage can stay nearly constant even though the gate voltage and consequently the source voltage swings up and down together by 20 or 30 volts or whatever. It is only the *difference* between gate and source voltage that is seen by the fet as bias and/or signal.
What really matters here is the gate-to-source voltage. Let's assume that at rest the g-s voltage is exactly 5v. Then, if the source swings from 0V to -10V, the gate would need to swing from +5v to -5v to let this happen. They have both gone down by 10v, so the gate is still 5v above the source. In reality it would be something like 4.9v above the source because the g-s voltage has to go down a little to reduce the fet conduction.
The g-s voltage can stay nearly constant even though the gate voltage and consequently the source voltage swings up and down together by 20 or 30 volts or whatever. It is only the *difference* between gate and source voltage that is seen by the fet as bias and/or signal.
Roll-your-own output choke.
I knew right from the start that the reason why this type of class A amp with it's output choke is not very common in DIY circles is because a suitable choke is sometimes hard to come by and/or not everyone has on in their junk box, let alone two. What's more, most diy'ers are a bit scared to have a go at winding their own. Now a mains transformer might be one thing, with safety issues and what-not, but a choke is pretty tame with it's single winding and no big deal on insulation for low voltage apps. But why use a choke anyway? Because it halves the power dissipation at idle. e.g if you want 50 watts output you will have 100 watts dissipation at standby, whereas if your amp has a constant current sink instead of a choke you will need double the dc supply rail at the same current so your 50 watt amp dumps 200 watts when all is quiet. You end up needing twice the number of power devices, double size power supply, double sized heatsink etc etc. Also, with a choke output amp, if one of your output devices decides to go s/c then it will momentarily pull a voltage across the choke until the fuse pops, the chokeless variety may well give the speaker voice coil a dose of DC trauma.
But I digress... I did a bit of "kerbside shopping" this afternoon and came home with two old microwave ovens. One was a Sanyo only several years old but with an arced magnetron.
Anyway, seeing these things are produced in the millions, the manufacturers investigate every method of simplifying construction to cut costs. One thing they cut corners on is the transformer, which judging by the size would be between 300-500 VA depending on the oven power. They actually get hammered quite a bit harder than this and have forced air cooling. To simplify automated assembly of the transformers, they are butt-stacked, that is, all the E laminations are stacked together and all the I laminations are also stacked together. They are plonked together and a seam weld is applied across the length of the joint. Normally transformers have their laminations interleaved with the E pieces stacked from alternate sides and the I pieces fill in the gaps. With the butt stacked method there is a slight air gap between the E stack and the I stack which would normally be undesirable but in an oven tranny the extra leakage reactance caused is actually desirable, sort of like in an arc welding transformer.
Taking a close look at one of my trannies, it occured to me that I could simply run an angle grinder along the length of the welded seam holding all the E's to the I's and then they would easily separate, and what's more, because of aditional welds, each would stay together as one piece instead of going into dozens of separate slices. Once the (one piece each!) E and I have been separated, it is a simple matter to slip off the original winding, also the magnetic shunt that fits between the primary and secondary windings. Throw that piece away. If your particular transformer has a bobbin then things will be easier from this point on. Remove the old wire from it. Wind up a new winding and slip it over the E centre leg and then replace the I piece, and put a layer of plastic between the E and I about 0.5mm thick to begin with. A suitable inductance would be about 100 milliHenries, the smaller the gap the higher the inductance (the higher the better actually) but it also makes the choke more easy to saturate, the magnetic equivalent of clipping.
What wire to use? My gut feeling is to fill the entire winding space using 1.5mm enamelled copper wire and begin with a 0.5mm magnetic airgap. See? Chokes are easy. Cheap too. I'll see if I can post some pics in the next couple of days chronicling my adventures with this approach.
I knew right from the start that the reason why this type of class A amp with it's output choke is not very common in DIY circles is because a suitable choke is sometimes hard to come by and/or not everyone has on in their junk box, let alone two. What's more, most diy'ers are a bit scared to have a go at winding their own. Now a mains transformer might be one thing, with safety issues and what-not, but a choke is pretty tame with it's single winding and no big deal on insulation for low voltage apps. But why use a choke anyway? Because it halves the power dissipation at idle. e.g if you want 50 watts output you will have 100 watts dissipation at standby, whereas if your amp has a constant current sink instead of a choke you will need double the dc supply rail at the same current so your 50 watt amp dumps 200 watts when all is quiet. You end up needing twice the number of power devices, double size power supply, double sized heatsink etc etc. Also, with a choke output amp, if one of your output devices decides to go s/c then it will momentarily pull a voltage across the choke until the fuse pops, the chokeless variety may well give the speaker voice coil a dose of DC trauma.
But I digress... I did a bit of "kerbside shopping" this afternoon and came home with two old microwave ovens. One was a Sanyo only several years old but with an arced magnetron.
Anyway, seeing these things are produced in the millions, the manufacturers investigate every method of simplifying construction to cut costs. One thing they cut corners on is the transformer, which judging by the size would be between 300-500 VA depending on the oven power. They actually get hammered quite a bit harder than this and have forced air cooling. To simplify automated assembly of the transformers, they are butt-stacked, that is, all the E laminations are stacked together and all the I laminations are also stacked together. They are plonked together and a seam weld is applied across the length of the joint. Normally transformers have their laminations interleaved with the E pieces stacked from alternate sides and the I pieces fill in the gaps. With the butt stacked method there is a slight air gap between the E stack and the I stack which would normally be undesirable but in an oven tranny the extra leakage reactance caused is actually desirable, sort of like in an arc welding transformer.
Taking a close look at one of my trannies, it occured to me that I could simply run an angle grinder along the length of the welded seam holding all the E's to the I's and then they would easily separate, and what's more, because of aditional welds, each would stay together as one piece instead of going into dozens of separate slices. Once the (one piece each!) E and I have been separated, it is a simple matter to slip off the original winding, also the magnetic shunt that fits between the primary and secondary windings. Throw that piece away. If your particular transformer has a bobbin then things will be easier from this point on. Remove the old wire from it. Wind up a new winding and slip it over the E centre leg and then replace the I piece, and put a layer of plastic between the E and I about 0.5mm thick to begin with. A suitable inductance would be about 100 milliHenries, the smaller the gap the higher the inductance (the higher the better actually) but it also makes the choke more easy to saturate, the magnetic equivalent of clipping.
What wire to use? My gut feeling is to fill the entire winding space using 1.5mm enamelled copper wire and begin with a 0.5mm magnetic airgap. See? Chokes are easy. Cheap too. I'll see if I can post some pics in the next couple of days chronicling my adventures with this approach.
Now we're getting somewhere. You can see where I ran the angle grinder disk along the welded seam so the two halves would separate. I cut in about 2 mm which is probably a little deep. About 1.5 mm would have been enough. The two faces were glued together as well so I had to hit it fairly hard with a hammer to get them apart. Be careful you don't break anything here.
Attachments
Once the two core halves are separated you then have to remove the coils off the centre pole. This one had a pair of wedges between the centre pole and the coils so that had to be tapped out with a screwdriver and hammer first. Then use the hammer to tap off the coils and you are there. Pic shows E's, I's, 240v winding, laminated iron shunts, high voltage winding, and magnetron filament winding. You need none of those, only the iron core. Next installment will be telling of my adventures actually rewinding the thing. 
Attachments
Different gaps for different chaps...
I wound 30 turns of wire on the core today and did a quickie measurement to get a rough idea of the core behaviour.
Just the E core only - 586 uH.
EI with (virtually) no gap - 3.54 mH (no good for DC but probably great for crossover choke - very low resistance because of few turns needed, & hi power).
0.75mm gap - 1.59mH
1.6 mm gap = 1 layer of pcb material - 1.161 mH
Inductance varies with the square of the turns, so doubling the turns would quadruple the inductance, tripling the turns would x9 the inductance etc.
Larger airgaps lower the inductance but allow it to carry more DC without saturating. A choke this physical size would be great for a class A amp with it's steady current drain.
I wound 30 turns of wire on the core today and did a quickie measurement to get a rough idea of the core behaviour.
Just the E core only - 586 uH.
EI with (virtually) no gap - 3.54 mH (no good for DC but probably great for crossover choke - very low resistance because of few turns needed, & hi power).
0.75mm gap - 1.59mH
1.6 mm gap = 1 layer of pcb material - 1.161 mH
Inductance varies with the square of the turns, so doubling the turns would quadruple the inductance, tripling the turns would x9 the inductance etc.
Larger airgaps lower the inductance but allow it to carry more DC without saturating. A choke this physical size would be great for a class A amp with it's steady current drain.
Howdy, gentlemen.
Could something like this project be done using a number of smaller chokes connected in series?
I recently bought two 25mH @ 3A (airgapped) double C-core chokes with 0,4 ohms Rdc and I think I know were i can get more of those. Would a series string of five per channel be enough?
If i buy ten more I could use two in the CLC PSU together with a 2x12 V 500VA toroid that I already have lying around.
Worth a try, you think??
Could something like this project be done using a number of smaller chokes connected in series?
I recently bought two 25mH @ 3A (airgapped) double C-core chokes with 0,4 ohms Rdc and I think I know were i can get more of those. Would a series string of five per channel be enough?
If i buy ten more I could use two in the CLC PSU together with a 2x12 V 500VA toroid that I already have lying around.
Worth a try, you think??
Yep, five should be heaps. 
With a total series resistance of 2 ohms for 5 series chokes, if you run say 3 amps dc quiescent current there will be a voltage drop of 6 volts of course, but this has an advantage in that the output fet current will be quite stable with temperature and so you will probably not need to have the dc feedback thingy to regulate the bias voltage.
With a total series resistance of 2 ohms for 5 series chokes, if you run say 3 amps dc quiescent current there will be a voltage drop of 6 volts of course, but this has an advantage in that the output fet current will be quite stable with temperature and so you will probably not need to have the dc feedback thingy to regulate the bias voltage.
Thanx for the reply!
I´ll place an order for ten chokes first thing tomorrow morning, together with a bunch of mosfets (3 or 4 IRF 540 per channel) and some miscellanous stuff. I can probably get a good price if I buy ten pieces of everything=)
Next step will be to locate cheap PSU caps and a pair of big heatsinks.
What about driving the whole thing with a nice little tube??
I´ll place an order for ten chokes first thing tomorrow morning, together with a bunch of mosfets (3 or 4 IRF 540 per channel) and some miscellanous stuff. I can probably get a good price if I buy ten pieces of everything=)
Next step will be to locate cheap PSU caps and a pair of big heatsinks.
What about driving the whole thing with a nice little tube??
I forgot to mention - if you have 6 volts drop across the output inductor(s) then you will definitely need a coupling cap between it and the loudspeaker. Also, the constant current source I put as a load above the voltage amplifier fet can be left out if you wan't; it's probably not that necessary and just adds complication. Instead, use a resistor that drops half the supply rail at 50mA. The bootstrapping of the top of the load resistor to that stage should stay there though.
Driven with a tube? Yum! Or do you mean just drive the output stage with a tube instead of a fet? If you mean drive the voltage amplifier stage with a tube instead of an opamp, note that it's input impedance is only 3k3 courtesy of the virtual earth presented by the voltage amplifier fet gate.
Driven with a tube?
Yum! Or do you mean just drive the output stage with a tube instead of a fet? If you mean drive the voltage amplifier stage with a tube instead of an opamp, note that it's input impedance is only 3k3 courtesy of the virtual earth presented by the voltage amplifier fet gate.
3x3300µF low ESR electrolytics bypassed with an 4µF paper in oil should do as output caps, don´t you think?
There will be paper in oils as PSU decoupling caps too, I have a good stock and they are just soooo much cooler than all those plastic film caps
I´m going to replase the whole VAS stage with a tube stage.
One ECC82 (12AU7) or 6SN7 per channel using one half of the tube as VAS and the other as cathode follower should give just about the right gain and output impedance I hope.
There will be paper in oils as PSU decoupling caps too, I have a good stock and they are just soooo much cooler than all those plastic film caps
I´m going to replase the whole VAS stage with a tube stage.
One ECC82 (12AU7) or 6SN7 per channel using one half of the tube as VAS and the other as cathode follower should give just about the right gain and output impedance I hope.
I´d love to post some schematics (and pictures of my earlier projects, by the way) but my scanner isn´t working and besides, I dont have a clue how to post pictures here
Anyways, I have ordered most of the parts and now I´m only hoping that the guy with the chokes has ten pieces left, otherwise things will become a bit complicated, to say the least.
The chokes are cheap surplus stuff and I dont have any idea what they are intended for and who has manufactured them.
Anyways, I have ordered most of the parts and now I´m only hoping that the guy with the chokes has ten pieces left, otherwise things will become a bit complicated, to say the least.
The chokes are cheap surplus stuff and I dont have any idea what they are intended for and who has manufactured them.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.