! jkeny said:
Assuming a minimum output of 20V PP and a 1:1 OPT, that would give you about 7V rms, so about 6 watts into 8 ohms. In the original Zeus, best distortion performance was with the 4:1 OPT, a little less with the 2:1. Don't know if she showed 1:1, but you might ask.
Sheldon
edit: Oops, forgot the negative rail myself. At 1:1 power would be about 25 watts, 6 watts for 2:1.
Sheldon said:
Yes this is true - but in spice experiments I found if you lower the power supply voltage and increase standing the current using the o/p transformer for reducing distortion seems to be unecessary - the transformer may have other advantages but I think distortion reduction is not one of them in these circumstances.
an o/p choke of course is still absolutely essential
mike
Thanks Mikelm and jkeny. This is interesting. I do plan to do a version of this amp when I finish a couple of other projects. Here is an interesting post which bears on the Zout requirements of the driving stage. Your thoughts? http://www.diyaudio.com/forums/showthread.php?postid=999010#post999010
Sheldon
Sheldon
suiraMB said:
Hmm, the more I think about this, the more interesting it is. In my previous post, I referenced a discussion about variation of Crss with output and its possible effect on phase related distortion. It looks like these FET's go from about 300pf at 20Vds to 600pf at 0Vds. That's a much lower ratio than the Mosfets I've looked at, especially as Vds approachs zero (and sometimes the steep change starts at 20 or 30Vds). The tube driver I have for this has a swing of 200V and a Zout of about 8K. So I could step than down 4:1 for a 25V swing and a Zout of 500, which would give a corner frequency of 500kHz. Even with a 2:1 step down on the OPT, I'd have 6 watts or so. Should be plenty for my application. If you are right about the distortion suiraMB, it should be no problem even going 1:1. I have little concern about the output impedence as I will be using these for driving only frequencies higher than about 300Hz. The only trick would seem to be designing the OPT to give the desired bias current. That should be possible from wire resistance tables. But, the bias power would have to be dissipated by the transformer. Wouldn't it be easier just to add some series resistance and get rid of the heat there?
Thanks again,
Sheldon
If I decide to go this way, and I'm leaning hard just because it seems very cool, I'd have no problem getting some extra's and sending them along. You can help me if I do, by pointing me to the post with the latest ordering and shipping info. I looked at the GB thread and it's a long one to dig through.
Sheldon
Sheldon
Sheldon,
Thanks, I would be delighted if you could do that.
As far as I know the info on the first post of GB thread is still current prices + shipping info http://www.diyaudio.com/forums/showthread.php?postid=814387#post814387
John
Thanks, I would be delighted if you could do that.
As far as I know the info on the first post of GB thread is still current prices + shipping info http://www.diyaudio.com/forums/showthread.php?postid=814387#post814387
John
jkeny said:
I'll let you know if I go this route, and I'll try to decide soon. No big gotcha's so far? And the simple biasing scheme is very attractive. I'm very new to ss device characteristics, but it looks like these FET's have a negative temp. coefficient, so, unlike the mosfet's, no special measures are required to avoid thermal runaway. Am I reading that right? If so, that's another nice feature.
Sheldon
Sheldon said:
Mosfets have a negative tempco.
Until you get them very too hot, then they give up the neg tempco idea and run away... In fact it is difficult to bias up Mosfets past a certain standing current unless you can keep them very very cool, they try to track back to where they want to sit...
_-_-bear
awprint
Hi All,
I trust everyone is keeping well.
I have been busy with work stuff so not so much time for audio these last few months.
Who - me?
I did look at these parts but as the standard mosfets are giving below 0.01% THD I didn't see the point of the additional complexity.
However I have used Supertex DN2535s for a Zeus PP line-driver
http://www.supertex.com/pdf/datasheets/DN2540.pdf
see some preliminary testing at:
http://www.audiophonics.com/audiophonics-zeus-line-driver-vmos-pp-1.html
They give good results.
Note however that the termination of X1's secondary has to be matched to the driving source impedance otherwise there can be significant HF peaking before rolloff.
This circuit always operates in Class A, so the auto bias method is okay. Voltage rating is 350 or 400 volts, so no problems here with a 24 volt supply.
The DN2535 has a max current of 500mA and 15W so would need to be paralleled up for an output stage driver. And I would not auto bias (unless always in Class A with a big step down ratio) so having to do a separate negative bias plus multiple devices somewhat increases the complexity and makes for a potential failure mode if the bias doesn't come up properly.
Using standard MOSFETS removal of the positive bias voltage effectively puts the amp into standby, which is a useful feature for energy savings when the amp is not in use.
Best wishes,
Susan.
I trust everyone is keeping well.
I have been busy with work stuff so not so much time for audio these last few months.
jkeny said:
Who - me?
wrenchone said:
I did look at these parts but as the standard mosfets are giving below 0.01% THD I didn't see the point of the additional complexity.
However I have used Supertex DN2535s for a Zeus PP line-driver
http://www.supertex.com/pdf/datasheets/DN2540.pdf
see some preliminary testing at:
http://www.audiophonics.com/audiophonics-zeus-line-driver-vmos-pp-1.html
They give good results.
Note however that the termination of X1's secondary has to be matched to the driving source impedance otherwise there can be significant HF peaking before rolloff.
This circuit always operates in Class A, so the auto bias method is okay. Voltage rating is 350 or 400 volts, so no problems here with a 24 volt supply.
The DN2535 has a max current of 500mA and 15W so would need to be paralleled up for an output stage driver. And I would not auto bias (unless always in Class A with a big step down ratio) so having to do a separate negative bias plus multiple devices somewhat increases the complexity and makes for a potential failure mode if the bias doesn't come up properly.
Using standard MOSFETS removal of the positive bias voltage effectively puts the amp into standby, which is a useful feature for energy savings when the amp is not in use.
Best wishes,
Susan.
Nice to see you here. I did try one channel with the regular Mosfets, the 801 driver, and an autoformer for the input. Got tunes out and sounded fine on a small speaker. I've got an RIAA project to finish first, but then I'll play some more with this one.
By added complexity, I assume you mean the cascode idea. I was thinking of trying them in a low power version as a self biased sub for a mosfet. As for the point, yes you are correct, but hey this is just for fun.
This amp is a little different to figure. Using the Lovaltech devices, what conditions would meet this requirement? I was looking at something on the order of 1amp bias current for a 10 volt supply and maybe a 2:1 output.
Sheldon
Susan-Parker said:
By added complexity, I assume you mean the cascode idea. I was thinking of trying them in a low power version as a self biased sub for a mosfet. As for the point, yes you are correct, but hey this is just for fun.
Susan-Parker said:
This amp is a little different to figure. Using the Lovaltech devices, what conditions would meet this requirement? I was looking at something on the order of 1amp bias current for a 10 volt supply and maybe a 2:1 output.
Sheldon
Hi,
Thanks. Please let us know how you get on.
Understood. And Agreed. This is hobby and DIY and meant to be relaxing etc. with personal enjoyment, satisfaction and learning.
I would probably look at using a standard MOSFET to give a reasonably regulated 11.75 volt supply as the extra couple of volts will make a big difference to the output watts possible.
Note from my VMOS schematic that the gate zeners are reversed in polarity. And here should probably be reduced to 9 volts.
Also don't forget the current capability of the transformer windings, as with these high biases there will be some heating.
One of the problems seems to me to be the package as it isn't overly big for the types of currents and wattages proposed.
I know that it is theoretically capable of 65 watts but once one has derated at higher temperatures e.g 80 degrees C it is only about 30 watts and the package has a thermal resistance junction to case of 1.8 degrees/watt - compared to 0.69 for TO247
But high bias with low voltage is certainly one way to go.
Per STW34NB20 I use 750mA at 34 volts supply in my standard power stage.
Will be interested to hear how this version fairs.
Best wishes,
Susan.
Sheldon said:
Thanks. Please let us know how you get on.
Understood. And Agreed. This is hobby and DIY and meant to be relaxing etc. with personal enjoyment, satisfaction and learning.
I would probably look at using a standard MOSFET to give a reasonably regulated 11.75 volt supply as the extra couple of volts will make a big difference to the output watts possible.
Note from my VMOS schematic that the gate zeners are reversed in polarity. And here should probably be reduced to 9 volts.
Also don't forget the current capability of the transformer windings, as with these high biases there will be some heating.
One of the problems seems to me to be the package as it isn't overly big for the types of currents and wattages proposed.
I know that it is theoretically capable of 65 watts but once one has derated at higher temperatures e.g 80 degrees C it is only about 30 watts and the package has a thermal resistance junction to case of 1.8 degrees/watt - compared to 0.69 for TO247
But high bias with low voltage is certainly one way to go.
Per STW34NB20 I use 750mA at 34 volts supply in my standard power stage.
Will be interested to hear how this version fairs.
Best wishes,
Susan.