Hi peranders,
I am not sure I follow you here.
My understanding is that cross conduction is where one device doesn't turn off fast enough and the complimentry device is still active and they then fight between themselves against the opposite power rails?
???
Best wishes,
Susan.
peranders said:
I am not sure I follow you here.
My understanding is that cross conduction is where one device doesn't turn off fast enough and the complimentry device is still active and they then fight between themselves against the opposite power rails?
???
Best wishes,
Susan.
Re: Re: Re: Confirmation on Affirmation
Either way it's a lot of power and standing on the ground looking up at the beasts I just SO know that there is a lot going on between those ears and if I do get to ride the thing it is only on it's sufference.
Best wishes,
Susan.
Currently listening to maybe 0.0015 of a horse.
peranders said:
... or 735.5 Watts if it's a metric one
Either way it's a lot of power and standing on the ground looking up at the beasts I just SO know that there is a lot going on between those ears and if I do get to ride the thing it is only on it's sufference.
Best wishes,
Susan.
Currently listening to maybe 0.0015 of a horse.
Re: Commentable Experienced Thoughts
Hi Amp_man,
Assuming you are using a proper 2:1 output transformer and not the center tapped inductor version I make that on the order of 741.125 watts.
Force air cooling - very good
Quiescent bias level set for about an amp total?
Exciting.
Thanks.
Best wishes,
Susan.
Hi Amp_man,
amp_man_1 said:
Assuming you are using a proper 2:1 output transformer and not the center tapped inductor version I make that on the order of 741.125 watts.
Force air cooling - very good
Quiescent bias level set for about an amp total?
Exciting.
Thanks.
Best wishes,
Susan.
Bass Speaker Drawings
Dear All,
For anyone who is interested about the delivery end of all this I have updated my bass speaker page and added 150 dpi resolution B/W drawings for printing (as requested).
http://www.susan-parker.co.uk/susan-speaker-bass.htm
There is also a link to the driver manufacturer's web site.
Only rated at about one foal though.
Best wishes,
Susan.
Dear All,
For anyone who is interested about the delivery end of all this I have updated my bass speaker page and added 150 dpi resolution B/W drawings for printing (as requested).
http://www.susan-parker.co.uk/susan-speaker-bass.htm
There is also a link to the driver manufacturer's web site.
Only rated at about one foal though.
Best wishes,
Susan.
IC Balanced Line Driver Input
Dear All,
Here are pics of the two stage amplifier driven using a SSM2142 Balanced Line Driver instead of the first transformer to drive the first stage mosfet gates.
http://www.susan-parker.co.uk/zeus-2stage-amp-test-ssm2142-drive-1.jpg
Full pic of the whole amp:
http://www.susan-parker.co.uk/zeus-2stage-amp-test-ssm2142-full-1.jpg
The SSM2142's GND input pin 3 is referenced to Vbias as is the Input pin 4 which has a 100k resistor to pin 3. The input signal is capacitively coupled to allow for the shift in levels.
Gain stage on this test board is a SSM2015, with 2 x 10uF input cap mixing down of stereo to mono for testing.
This should also work with the DRV134 parts as they look to be the same configuration (and same pin numbers).
Best wishes,
Susan.
Dear All,
Here are pics of the two stage amplifier driven using a SSM2142 Balanced Line Driver instead of the first transformer to drive the first stage mosfet gates.
http://www.susan-parker.co.uk/zeus-2stage-amp-test-ssm2142-drive-1.jpg
Full pic of the whole amp:
http://www.susan-parker.co.uk/zeus-2stage-amp-test-ssm2142-full-1.jpg
The SSM2142's GND input pin 3 is referenced to Vbias as is the Input pin 4 which has a 100k resistor to pin 3. The input signal is capacitively coupled to allow for the shift in levels.
Gain stage on this test board is a SSM2015, with 2 x 10uF input cap mixing down of stereo to mono for testing.
This should also work with the DRV134 parts as they look to be the same configuration (and same pin numbers).
Best wishes,
Susan.
Re: A little more transformer info....
Yep, here's what Sowter's sales rep emailed:
Thomas & Skinner has three thicknesses of M-6 Orthosil laminations - 004", .006" and .014" or for those in the metric world - 0.1016mm , 0.1524mm, and 0.3556mm
How tough do you think it would be to work with the two thinnest lams? They sound almost like tissue paper to me.
I don't quite follow the "running" the transformers at 2:1 or 4:1 part. Does that refer to the winding ratio?
On the 200VA toroid you suggest a single "Quad-filar wind of 0.8 mm enameled magnet wire w/ 120 to 150 turns." But, on the E-I, there are primary and secondary windings.
Could you please clarify this for little ol' me? I'm a bit confused.
Susan-Parker said:
Yep, here's what Sowter's sales rep emailed:
Susan-Parker said:
Thomas & Skinner has three thicknesses of M-6 Orthosil laminations - 004", .006" and .014" or for those in the metric world - 0.1016mm , 0.1524mm, and 0.3556mm
How tough do you think it would be to work with the two thinnest lams? They sound almost like tissue paper to me.
I don't quite follow the "running" the transformers at 2:1 or 4:1 part. Does that refer to the winding ratio?
On the 200VA toroid you suggest a single "Quad-filar wind of 0.8 mm enameled magnet wire w/ 120 to 150 turns." But, on the E-I, there are primary and secondary windings.
Could you please clarify this for little ol' me? I'm a bit confused.
Re: IC Balanced Line Driver Input
Very nice. More importantly, very cheap
How does it sound compared to using an input transformer?
You stated in an earlier post that:
Was that because of too much gain? input trans (1:10) + 6 dB from the SSM2142.
Or, was that a matter of just sounding terrible?
How does it work gain-wise? The input trans you've been using are 1:9 or 1:10, the SSM2142/SSM2015 combo is 21dB, I think (1:5?).
The SSM2142 provides 6dB gain. I can't find a datasheet for the SSM2015 anywhere, including AD. But, it is mentioned in the SSM2402/
SSM2412 datasheet (pg. 9, Fig 7) as being set to a gain of 15dB.
It's too bad the balanced line receivers I've seen so far(INA134/137, SSM2015) convert the differential single back to single-ended. It would be great to use the Pro Audio approach of a line amplifier at the preamp and receiver in the Zeus in lieu of the input tranformer.
We'd get all the advantages of noise rejection and the ability to drive any length cables with extremely low distortion.
Do you think it would be possible to feed the + and - differential signals to seperate receivers in the INA2137 dual receiver pictured below and still maintain balance?
The + input for both receivers would share the common ground line of the differential cable.
Anyway, just a thought.
Susan-Parker said:
Very nice. More importantly, very cheap
How does it sound compared to using an input transformer?
You stated in an earlier post that:
Was that because of too much gain? input trans (1:10) + 6 dB from the SSM2142.
Or, was that a matter of just sounding terrible?
How does it work gain-wise? The input trans you've been using are 1:9 or 1:10, the SSM2142/SSM2015 combo is 21dB, I think (1:5?).
The SSM2142 provides 6dB gain. I can't find a datasheet for the SSM2015 anywhere, including AD. But, it is mentioned in the SSM2402/
SSM2412 datasheet (pg. 9, Fig 7) as being set to a gain of 15dB.
It's too bad the balanced line receivers I've seen so far(INA134/137, SSM2015) convert the differential single back to single-ended. It would be great to use the Pro Audio approach of a line amplifier at the preamp and receiver in the Zeus in lieu of the input tranformer.
We'd get all the advantages of noise rejection and the ability to drive any length cables with extremely low distortion.
Do you think it would be possible to feed the + and - differential signals to seperate receivers in the INA2137 dual receiver pictured below and still maintain balance?
An externally hosted image should be here but it was not working when we last tested it.
The + input for both receivers would share the common ground line of the differential cable.
Anyway, just a thought.
Re: Re: A little more transformer info....
Hi Darkmoebius
Ah, great. Thanks for that confirmation.
The M6 I have been using is the 0.35 mm / 0.014" thickness material.
These thinner laminations I would call it shim, but yes it is getting down to tissue paper thickness.
The 0.006" / 0.15 mm is the best compromise between manageability and at under half the thickness of the standard laminations should give much improved theoretical magnetization curves. Going to the even finer laminations is only a third thinner than this, and probably not worth the extra hassle.
These laminations can be very sharp and care should be taken when handling not to get the equivalent of paper cuts (how do I know this?). I would recommend wearing light cotton gloves to keep the individual laminations as clean as possible.
Remember when assembling the frames/shrouds to go round the bolts a few times sequentially tightening as you want the stack well compressed (the same as tightening the wheel bolts on a car after changing a wheel).
I can't say that it is necessarily going to be that audible or easy to measure the differences without very careful A/B comparisons as there are so many factors influencing the final sound quality.
But for the additional effort of building these transformers oneself one might as well be happy that one has the best configuration.
FYI I am intending to switch over to using the 0.006" / 0.15 mm laminations once I have used up my current stock.
Okay.
Example 1:
If you use a power mains toroid transformer with bifilar wound secondaries and, using these secondaries only, wire the speaker directly across the windings you have a 1:1 coupling from the voltage across the mosfet's sources to the speaker.
Example 2:
If you quad filar wind the transformer then take two windings and wire them in series with the center grounded and the two out of phase ends to the mosfet's sources, then wire the other two separately in parallel for the speaker, you have a 2:1 step down transformer.
Example 3:
Quad filar wind the transformer as before, but split two of the windings half way.
Take the two full windings and wire them in series with the center grounded and the two out of phase ends to the mosfet's sources.
Wire the other two sets of four windings into two pairs of parallel windings and connect to a switch.
Set the switch to series for the speaker you have a 2:1 step down transformer, set the switch to parallel for the speaker you have a 4:1 step down transformer.
==
For the 75 watt transformer (new spec N size that Sowter will be making) in standard transformer configuration there are four output windings and one could bring them out to an 8 way Neutrik Speakon style connector for maximum flexibility. Then it is just a plug wiring configuration.
For a hard wired configuration the Multi-Contact 2 Powerline 6mm series looks interesting as an alternative for higher powers and larger cable diameters.
http://www.multi-contact.com/
"Round Connectors, up to 600V, up to 125A, Ø 6mm, single-pole, insulated."
I have a set of these which are going to be used in my bass driver amp.
Model ID/B6AR-N-S safety panel receptacles (sockets - chassis mount with bolt cable fixing).
14.0010-21
14.0010-22
http://www.multi-contact.com/Products/IS/2_Powerline/14.0010 ID_B6AR-N-S.gif
Model KST6AR-N/... plugs
15.5001-21
15.5001-22
http://www.multi-contact.com/Products/IS/2_Powerline/15.0001 KST6AR-N_10.gif
=====
The reason for having these options (Example 3) is to better match the loudspeaker load to the amplifier output.
From the previously posted distortion figures one can see that distortion rises as the load is decreased in impedance (I will be adding these to my website shortly).
By switching around the four output windings one can match to the get better distortion characteristics.
Going from 2:1 to 4:1 step down ratios will reduce the distortion figures for the same load impedance - I have measured down to a quarter going from 8 to 16 ohms.
Obviously the amount of power is reduced but this is part of the trade off.
I hope this helps.
Best wishes,
Susan.
Hi Darkmoebius
darkmoebius said:
Ah, great. Thanks for that confirmation.
The M6 I have been using is the 0.35 mm / 0.014" thickness material.
These thinner laminations I would call it shim, but yes it is getting down to tissue paper thickness.
The 0.006" / 0.15 mm is the best compromise between manageability and at under half the thickness of the standard laminations should give much improved theoretical magnetization curves. Going to the even finer laminations is only a third thinner than this, and probably not worth the extra hassle.
These laminations can be very sharp and care should be taken when handling not to get the equivalent of paper cuts (how do I know this?). I would recommend wearing light cotton gloves to keep the individual laminations as clean as possible.
Remember when assembling the frames/shrouds to go round the bolts a few times sequentially tightening as you want the stack well compressed (the same as tightening the wheel bolts on a car after changing a wheel).
I can't say that it is necessarily going to be that audible or easy to measure the differences without very careful A/B comparisons as there are so many factors influencing the final sound quality.
But for the additional effort of building these transformers oneself one might as well be happy that one has the best configuration.
FYI I am intending to switch over to using the 0.006" / 0.15 mm laminations once I have used up my current stock.
Okay.
Example 1:
If you use a power mains toroid transformer with bifilar wound secondaries and, using these secondaries only, wire the speaker directly across the windings you have a 1:1 coupling from the voltage across the mosfet's sources to the speaker.
Example 2:
If you quad filar wind the transformer then take two windings and wire them in series with the center grounded and the two out of phase ends to the mosfet's sources, then wire the other two separately in parallel for the speaker, you have a 2:1 step down transformer.
Example 3:
Quad filar wind the transformer as before, but split two of the windings half way.
Take the two full windings and wire them in series with the center grounded and the two out of phase ends to the mosfet's sources.
Wire the other two sets of four windings into two pairs of parallel windings and connect to a switch.
Set the switch to series for the speaker you have a 2:1 step down transformer, set the switch to parallel for the speaker you have a 4:1 step down transformer.
==
For the 75 watt transformer (new spec N size that Sowter will be making) in standard transformer configuration there are four output windings and one could bring them out to an 8 way Neutrik Speakon style connector for maximum flexibility. Then it is just a plug wiring configuration.
For a hard wired configuration the Multi-Contact 2 Powerline 6mm series looks interesting as an alternative for higher powers and larger cable diameters.
http://www.multi-contact.com/
"Round Connectors, up to 600V, up to 125A, Ø 6mm, single-pole, insulated."
I have a set of these which are going to be used in my bass driver amp.
Model ID/B6AR-N-S safety panel receptacles (sockets - chassis mount with bolt cable fixing).
14.0010-21
14.0010-22
http://www.multi-contact.com/Products/IS/2_Powerline/14.0010 ID_B6AR-N-S.gif
Model KST6AR-N/... plugs
15.5001-21
15.5001-22
http://www.multi-contact.com/Products/IS/2_Powerline/15.0001 KST6AR-N_10.gif
=====
The reason for having these options (Example 3) is to better match the loudspeaker load to the amplifier output.
From the previously posted distortion figures one can see that distortion rises as the load is decreased in impedance (I will be adding these to my website shortly).
By switching around the four output windings one can match to the get better distortion characteristics.
Going from 2:1 to 4:1 step down ratios will reduce the distortion figures for the same load impedance - I have measured down to a quarter going from 8 to 16 ohms.
Obviously the amount of power is reduced but this is part of the trade off.
I hope this helps.
Best wishes,
Susan.
Re: Re: IC Balanced Line Driver Input
Hi darkmoebius
Sound is okay, but as I am not able to do an A/B comparison it is difficult to tell as auditory memory is (in my case anyway) quite fickle.
What you see in the picture of the full amp laid out on my desk is almost all the space I have to work on. I literally only have 24" x 16" desk space, which is mainly taken up with keyboard and mouse occupancy when I am using my computer (which makes a fair amount of noise).
But I have not gone "Aagh!" and have been listening to CDs all morning which is unusual for me as I normally prefer to work in silence.
Input transformer - preferably driven straight from the DAC current outputs - is of course the preferred implementation.
Just didn't work at any sort of drive level into the transformer. The waveforms were all over the place when I tried a sine sweep. Didn't get to the point of listening.
This method of driving the gates with the SSM2142 provides an alternative solution to the TI chip (very nice, very fiddly) for the low toroid impedances which are circa 15 ohms in my case.
BUT this configuration isn't recommended with the Sowter style input transformer as that wont have cross coupled (i.e.bifilar) windings.
From experience with the TI TPA6120 headphone chip I would now try bringing back the center tap of a 600 ohm impedance series connected input transformer and referencing it to the GND pin on the SSM2142. This MIGHT solve that problem.
One point for me (which I appreciate is not the same issues for most people following this thread) is the whole business of compliance testing and selling product. Whilst I am not doing this for the Zeus amplifier it is still something that I cling to as a matter of professional principle (probably misguided - but we all have our little foibles).
With the power amplifier as a mono-block with transformer linear power supply with a small storage capacitor I get low power line distortion and thus don't need to do power factor correction. Running under 50 volts DC for the power rail I am also in a lower safety category.
With transformer coupled inputs and outputs I have a fully isolated "box" and therefor much of the RFI/EMC and ESD testing is significantly simplified.
http://retards.org/library/technology/electronics/audio_circuits/synths/components/ssm/
has scans of data sheets for the SSM parts
In the test circuit (which is a heavily modified Maplin SSM2015 mic preamp kit) I have fiddled the gain to reduce it down somewhat.
Yes, but perhaps not the way you have in mind.
http://focus.ti.com/lit/ds/symlink/ina2137.pdf
referring to Figure 2.
Wire one up as per diagram and the second with the input pins swapped over. I.e. -in amp 1 to +in amp 2, and +in amp 1 to -in amp 2. This gives you a balanced output. Don't forget to put normalized 600 ohm resistor across the input.
Do NOT refer the input signal to the common ground line which should go to the power supply common (probably via a switch to isolate if there are ground loop issues).
Some back to back zeners as well perhaps?
However you still have to offset the balanced output to get the bias and without a bit more thought I am not sure that this can be done directly at this point.
There is a problem with running end to end with balanced systems where the two signals are not referred to a common point at some stage. I can't remember the exact details but there is an issue lurking somewhere.
Best wishes,
Susan.
Hi darkmoebius
darkmoebius said:Very nice. More importantly, very cheap
How does it sound compared to using an input transformer?
Sound is okay, but as I am not able to do an A/B comparison it is difficult to tell as auditory memory is (in my case anyway) quite fickle.
What you see in the picture of the full amp laid out on my desk is almost all the space I have to work on. I literally only have 24" x 16" desk space, which is mainly taken up with keyboard and mouse occupancy when I am using my computer (which makes a fair amount of noise).
But I have not gone "Aagh!" and have been listening to CDs all morning which is unusual for me as I normally prefer to work in silence.
Input transformer - preferably driven straight from the DAC current outputs - is of course the preferred implementation.
Just didn't work at any sort of drive level into the transformer. The waveforms were all over the place when I tried a sine sweep. Didn't get to the point of listening.
This method of driving the gates with the SSM2142 provides an alternative solution to the TI chip (very nice, very fiddly) for the low toroid impedances which are circa 15 ohms in my case.
BUT this configuration isn't recommended with the Sowter style input transformer as that wont have cross coupled (i.e.bifilar) windings.
From experience with the TI TPA6120 headphone chip I would now try bringing back the center tap of a 600 ohm impedance series connected input transformer and referencing it to the GND pin on the SSM2142. This MIGHT solve that problem.
One point for me (which I appreciate is not the same issues for most people following this thread) is the whole business of compliance testing and selling product. Whilst I am not doing this for the Zeus amplifier it is still something that I cling to as a matter of professional principle (probably misguided - but we all have our little foibles).
With the power amplifier as a mono-block with transformer linear power supply with a small storage capacitor I get low power line distortion and thus don't need to do power factor correction. Running under 50 volts DC for the power rail I am also in a lower safety category.
With transformer coupled inputs and outputs I have a fully isolated "box" and therefor much of the RFI/EMC and ESD testing is significantly simplified.
http://retards.org/library/technology/electronics/audio_circuits/synths/components/ssm/
has scans of data sheets for the SSM parts
In the test circuit (which is a heavily modified Maplin SSM2015 mic preamp kit) I have fiddled the gain to reduce it down somewhat.
Yes, but perhaps not the way you have in mind.
http://focus.ti.com/lit/ds/symlink/ina2137.pdf
referring to Figure 2.
Wire one up as per diagram and the second with the input pins swapped over. I.e. -in amp 1 to +in amp 2, and +in amp 1 to -in amp 2. This gives you a balanced output. Don't forget to put normalized 600 ohm resistor across the input.
Do NOT refer the input signal to the common ground line which should go to the power supply common (probably via a switch to isolate if there are ground loop issues).
Some back to back zeners as well perhaps?
However you still have to offset the balanced output to get the bias and without a bit more thought I am not sure that this can be done directly at this point.
There is a problem with running end to end with balanced systems where the two signals are not referred to a common point at some stage. I can't remember the exact details but there is an issue lurking somewhere.
Best wishes,
Susan.
Confirmation on Affirmation
Hi Susan- parker
We are going to drive the mosfets without inptut trafo,
we want to use balanced input signal of 15V to drive APT's
and we will also use an Symmetrical autotransformer for output stage to get maximum hi fi response with 2KW core.
Regards
Workhorse Technologies
Hi Susan- parker
We are going to drive the mosfets without inptut trafo,
we want to use balanced input signal of 15V to drive APT's
and we will also use an Symmetrical autotransformer for output stage to get maximum hi fi response with 2KW core.
Regards
Workhorse Technologies
Re: IC Balanced Line Driver Input
Same here. I tried cleaning up my desk this weekend and it only yielded an extra foot of room. My only solution is to literally move my bed to the other side of the room and bring in a foldup table to extend my desk along the longest wall. I may need to practice the long jump to get into bed
Can it get anymore elegant and simple than that?
Unfortunately, my DAC has single-ended outputs, so I have to add a differential amp or line driver to the outputs. Luckily, that's not too difficult since it's a DIY kit and I'm going to use a chassis w/ lots of room to grow.
BTW, have you figured out the approximate stepup ratio of the 15VA input toroid?
As we discussed earlier, the AD815 datasheet displays a standalone SE->diff preamp(Figure 53, datasheet) to directly drive a transformer and Gain can easily be varied from -1 to +10 by altering the ratio of the feedback resistor and input/gain resistor(Table I, datasheet).
Carlosfm had a thread in September comparing the AD815, TPA6120A2, and his longtime favorite OPA627+BUF634 as standalone SE preamps. For those interested, here's his schematic and description of the circuit.
For those who want the punchline:
Not that I think this makes it the best, but that at least someone has gotten satisfactory results with the chip. And since I have a sample pair sitting on my desk right now, I'll try them first.
The only downside is that Analog Devices only carries it in 24 pin SOIC and these things are ridiculously tiny!! I can't see any way to solder the leads.
Susan-Parker said:
Same here. I tried cleaning up my desk this weekend and it only yielded an extra foot of room. My only solution is to literally move my bed to the other side of the room and bring in a foldup table to extend my desk along the longest wall. I may need to practice the long jump to get into bed
Can it get anymore elegant and simple than that?
Unfortunately, my DAC has single-ended outputs, so I have to add a differential amp or line driver to the outputs. Luckily, that's not too difficult since it's a DIY kit and I'm going to use a chassis w/ lots of room to grow.
BTW, have you figured out the approximate stepup ratio of the 15VA input toroid?
As we discussed earlier, the AD815 datasheet displays a standalone SE->diff preamp(Figure 53, datasheet) to directly drive a transformer and Gain can easily be varied from -1 to +10 by altering the ratio of the feedback resistor and input/gain resistor(Table I, datasheet).
Carlosfm had a thread in September comparing the AD815, TPA6120A2, and his longtime favorite OPA627+BUF634 as standalone SE preamps. For those interested, here's his schematic and description of the circuit.
For those who want the punchline:
Not that I think this makes it the best, but that at least someone has gotten satisfactory results with the chip. And since I have a sample pair sitting on my desk right now, I'll try them first.
The only downside is that Analog Devices only carries it in 24 pin SOIC and these things are ridiculously tiny!! I can't see any way to solder the leads.
Re: Re: IC Balanced Line Driver Input
Hi darkmoebius,
Something like that for me as well but I think this might not meet with approval from certain other parties.
Um, see later...
I am using a 1+1:10+10 (ish) with a 12+12:115+115 mains toroid running backwards. This gives me a 1:10 step up.
The input impedance in series is about 60 ohms, which is more than the simple op-amp chips can handle. This is with the 3K7 Rterm necessary to get a flat response.
However the AD815 should be able to handle this.
I looked at the data sheet and said "Oops!" To my charign I have discovered that I actually have one of these which I bought in a fit of enthusiasm about four years ago but didn't get any further with trying it out. I even have a matching heatsink for it.
Please note that the data sheet is incorrect and with +- 15 volt supplies the AD815 can drive 20 volts PP NOT 40 as stated. That's the voltage the other side of a 1:2 step up transformer.
The schematic on the front page is also wrong. Whoever did the data sheet must have been having a bad hair day.
Personally I think you could use a DRV134 to split the phases and the simpler circuit as shown on the front page - fewer expensive precision resistors needed.
At the end of the day the input transformer ratio is dependent upon the supply rail of the power amp and the originating input signal level.
Assume 34 volt DC power for the amplifier, and allowing for the fact that although one can drive into the power rails it doesn't benefit one sonically, so thats about 60 volts PP input drive although allowing for overheads etc. one could figure on 80 volts PP.
1 volt PP input signal to the system.
DRV134 gives a gain of 2 - so we have a 2 volt PP output.
AD815 as show with a gain of 10 gives us 20 volts PP ( which is it's max output).
Toroid therefor only needs a gain of four times to give us our 80 volts PP.
Highest windings I have for a 15VA toroid (from RS) is 2 x 18 volts, for a 30 VA toroid is 2 x 25.
36 volts to 230 volts is a 6.4 times step up.
50 volts to 230 volts is a 4.6 times step up.
One can do various juggles with the numbers - you might want to run with a higher power supply rail for example.
The lower the step up ratio the higher the impedance one has to drive, and the lower the "reflected" mosfet gate capacitance.
Most of the pins are n/c which helps
I use standard 0.1" matrix board and lift every second pin so they come straight out and wire to these directly. Don't fold up the other way as they might break off.
PTH micro board is good as one can blob stuff together to make a heatsink.
The AD815 is a current feedback amp - so you should be able to get nice flat square waves for those who like to see such things
Hope this helps.
Best wishes,
Susan.
Hi darkmoebius,
darkmoebius said:
Same here. I tried cleaning up my desk this weekend and it only yielded an extra foot of room. My only solution is to literally move my bed to the other side of the room and bring in a foldup table to extend my desk along the longest wall. I may need to practice the long jump to get into bed
Something like that for me as well but I think this might not meet with approval from certain other parties.
Um, see later...
I am using a 1+1:10+10 (ish) with a 12+12:115+115 mains toroid running backwards. This gives me a 1:10 step up.
The input impedance in series is about 60 ohms, which is more than the simple op-amp chips can handle. This is with the 3K7 Rterm necessary to get a flat response.
However the AD815 should be able to handle this.
I looked at the data sheet and said "Oops!" To my charign I have discovered that I actually have one of these which I bought in a fit of enthusiasm about four years ago but didn't get any further with trying it out. I even have a matching heatsink for it.
Please note that the data sheet is incorrect and with +- 15 volt supplies the AD815 can drive 20 volts PP NOT 40 as stated. That's the voltage the other side of a 1:2 step up transformer.
The schematic on the front page is also wrong. Whoever did the data sheet must have been having a bad hair day.
Personally I think you could use a DRV134 to split the phases and the simpler circuit as shown on the front page - fewer expensive precision resistors needed.
At the end of the day the input transformer ratio is dependent upon the supply rail of the power amp and the originating input signal level.
Assume 34 volt DC power for the amplifier, and allowing for the fact that although one can drive into the power rails it doesn't benefit one sonically, so thats about 60 volts PP input drive although allowing for overheads etc. one could figure on 80 volts PP.
1 volt PP input signal to the system.
DRV134 gives a gain of 2 - so we have a 2 volt PP output.
AD815 as show with a gain of 10 gives us 20 volts PP ( which is it's max output).
Toroid therefor only needs a gain of four times to give us our 80 volts PP.
Highest windings I have for a 15VA toroid (from RS) is 2 x 18 volts, for a 30 VA toroid is 2 x 25.
36 volts to 230 volts is a 6.4 times step up.
50 volts to 230 volts is a 4.6 times step up.
One can do various juggles with the numbers - you might want to run with a higher power supply rail for example.
The lower the step up ratio the higher the impedance one has to drive, and the lower the "reflected" mosfet gate capacitance.
As we discussed earlier, the AD815 datasheet displays a standalone SE->diff preamp(Figure 53, datasheet) to directly drive a transformer and Gain can easily be varied from -1 to +10 by altering the ratio of the feedback resistor and input/gain resistor(Table I, datasheet).
Not that I think this makes it the best, but that at least someone has gotten satisfactory results with the chip. And since I have a sample pair sitting on my desk right now, I'll try them first.
The only downside is that Analog Devices only carries it in 24 pin SOIC and these things are ridiculously tiny!! I can't see any way to solder the leads.
Most of the pins are n/c which helps
I use standard 0.1" matrix board and lift every second pin so they come straight out and wire to these directly. Don't fold up the other way as they might break off.
PTH micro board is good as one can blob stuff together to make a heatsink.
The AD815 is a current feedback amp - so you should be able to get nice flat square waves for those who like to see such things
Hope this helps.
Best wishes,
Susan.
STW34NB20 mosfets
Dear All,
I hope everyone is keeping well.
Some further work on the amplifier with the Sowter input transformer and the 75 watt output transformer (similar to the spec that I have given to Sowter, details of which are on my website) was prompted by finding some STW34NB20 PowerMESH mosfets whilst digging around in some boxes looking for something else
See: http://www.st.com/stonline/books/pdf/docs/5407.pdf for the data sheet.
These STW34NB20 PowerMESH mosfets look good from initial testing, even though I only have three to hand so even with swapping around I didn't have very good matching.
(The closer matched pair does show better results so matching would seem to be a worthwhile exercise.)
Even so distortion levels are lower than with the hexfets for similar biasing levels and are noticeably improved when the bias is increased.
Supply = 34 volts
Quiescent current = 1.5 amps (for the pair).
Vbias = 5.24 volts
Q1 = 545 mV
Q2 = 619 mV
N.B. The 300 x 150 x 40 mm 0·36°C/W heatsink gets hot!
On a quick cursory test this morning even with this poor matching I have distortion levels well below 0.2% between 1 and 8 watts levels into 8 ohms.
I am using the sound card to generate the sine waves for these tests and using the digital attenuator. Unfortunately the distortion inherently raises as the levels are dropped so I need to get a separate level control back into the analog part of the circuit as I can't tell at the moment whether this is from bias mismatch or digital quantization (I suspect the latter).
SoundCard > TPA6120 > Sowter 1:5+5 > EI120 2:1 OutTx > 8 ohm load > HP8903B
I also did a frequency sweep at 6.5 watts between 20 Hz to 20 kHz.
Hz % Dist
20 0.641
25 0.447
30 0.331
35 0.265
40 0.220
45 0.192
50 0.174
60 0.151
70 0.140
80 0.134
90 0.131
100 0.129
125 0.125
150 0.125
175 0.124
200 0.124
250 0.124
300 0.124
400 0.125
500 0.125
600 0.126
700 0.128
800 0.130
900 0.132
1000 0.135
1100 0.137
1200 0.140
1300 0.142
1400 0.144
1500 0.148
1700 0.156
2000 0.166
2500 0.184
3000 0.218
4000 0.240
5000 0.274
7000 0.332
10000 0.393
12000 0.421
15000 0.459
20000 0.517
Again the rise in distortion levels at the higher frequencies may be soundcard or amp, and I can't tell just at the moment.
=====
Please note that due to the lower Crss capacitance of the mosfet the Gate resistors are increased to 470 ohms to avoid parasitic oscillations.
Even without a properly matched set I have better performance that that which I have had so far with the IRF140/150s, so these do look promising.
They will need matching as from the very detailed statistical analysis of my sample of three the spread seems wide between devices of same batch number (but perhaps different wafers?).
Available from DigiKey at UKP 1.58 each (10+ price).
So not too expensive to buy a handful and sort out.
I will have some matched 2SK1529 lateral FETs soon. It will be interesting to see how these perform in comparison.
Best wishes,
Susan.
Dear All,
I hope everyone is keeping well.
Some further work on the amplifier with the Sowter input transformer and the 75 watt output transformer (similar to the spec that I have given to Sowter, details of which are on my website) was prompted by finding some STW34NB20 PowerMESH mosfets whilst digging around in some boxes looking for something else
See: http://www.st.com/stonline/books/pdf/docs/5407.pdf for the data sheet.
These STW34NB20 PowerMESH mosfets look good from initial testing, even though I only have three to hand so even with swapping around I didn't have very good matching.
(The closer matched pair does show better results so matching would seem to be a worthwhile exercise.)
Even so distortion levels are lower than with the hexfets for similar biasing levels and are noticeably improved when the bias is increased.
Supply = 34 volts
Quiescent current = 1.5 amps (for the pair).
Vbias = 5.24 volts
Q1 = 545 mV
Q2 = 619 mV
N.B. The 300 x 150 x 40 mm 0·36°C/W heatsink gets hot!
On a quick cursory test this morning even with this poor matching I have distortion levels well below 0.2% between 1 and 8 watts levels into 8 ohms.
I am using the sound card to generate the sine waves for these tests and using the digital attenuator. Unfortunately the distortion inherently raises as the levels are dropped so I need to get a separate level control back into the analog part of the circuit as I can't tell at the moment whether this is from bias mismatch or digital quantization (I suspect the latter).
SoundCard > TPA6120 > Sowter 1:5+5 > EI120 2:1 OutTx > 8 ohm load > HP8903B
I also did a frequency sweep at 6.5 watts between 20 Hz to 20 kHz.
Hz % Dist
20 0.641
25 0.447
30 0.331
35 0.265
40 0.220
45 0.192
50 0.174
60 0.151
70 0.140
80 0.134
90 0.131
100 0.129
125 0.125
150 0.125
175 0.124
200 0.124
250 0.124
300 0.124
400 0.125
500 0.125
600 0.126
700 0.128
800 0.130
900 0.132
1000 0.135
1100 0.137
1200 0.140
1300 0.142
1400 0.144
1500 0.148
1700 0.156
2000 0.166
2500 0.184
3000 0.218
4000 0.240
5000 0.274
7000 0.332
10000 0.393
12000 0.421
15000 0.459
20000 0.517
Again the rise in distortion levels at the higher frequencies may be soundcard or amp, and I can't tell just at the moment.
=====
Please note that due to the lower Crss capacitance of the mosfet the Gate resistors are increased to 470 ohms to avoid parasitic oscillations.
Even without a properly matched set I have better performance that that which I have had so far with the IRF140/150s, so these do look promising.
They will need matching as from the very detailed statistical analysis of my sample of three the spread seems wide between devices of same batch number (but perhaps different wafers?).
Available from DigiKey at UKP 1.58 each (10+ price).
So not too expensive to buy a handful and sort out.
I will have some matched 2SK1529 lateral FETs soon. It will be interesting to see how these perform in comparison.
Best wishes,
Susan.
Hi PMA,
Thank you for your post.
Yes, you are quite correct.
I have a wonderful E-mu 1212 soundcard with balanced inputs and outputs - amazing specifications, distortion figures etc. etc. But...
... it sits inside a Pentium-4 3 GHz computer and runs under WinXP.
And there are an awful lot of software drivers, application interfaces, dynamic buffers and other stuff that has to work just right.
When it all works properly, it's great.
When it doesn't, one is up a gum tree.
Which is why I spend a thousand US dollars some time ago on a HP 8903B Audio analyser which gives me an independant and more importantly calibrated standard to work from.
I had some trouble with it's signal generator side of things which has been a real pain, but I managed to get it working again this afternoon (I think it got inadvertantly set into remote GPIB mode).
More and better results are forthcoming in a little while - but I have to go and pour the tea.
Best wishes,
Susan.
Thank you for your post.
PMA said:
Yes, you are quite correct.
I have a wonderful E-mu 1212 soundcard with balanced inputs and outputs - amazing specifications, distortion figures etc. etc. But...
... it sits inside a Pentium-4 3 GHz computer and runs under WinXP.
And there are an awful lot of software drivers, application interfaces, dynamic buffers and other stuff that has to work just right.
When it all works properly, it's great.
When it doesn't, one is up a gum tree.
Which is why I spend a thousand US dollars some time ago on a HP 8903B Audio analyser which gives me an independant and more importantly calibrated standard to work from.
I had some trouble with it's signal generator side of things which has been a real pain, but I managed to get it working again this afternoon (I think it got inadvertantly set into remote GPIB mode).
More and better results are forthcoming in a little while - but I have to go and pour the tea.
Best wishes,
Susan.