I'm thinking of using some 2SJ79/2SK216 lateral mosfets for the drivers, possibly fast and might smooth out the sound a bit. I see that AmpsLab is doing this on one of their amps.
http://www.ampslab.com/bi70mk2.htm
Might be interesting and I have these on hand.
http://www.ampslab.com/bi70mk2.htm
Might be interesting and I have these on hand.
will only one driver connected directly to an output pair have sufficient current gain?
It feels like that to utilize the whole voltage range of the amp (and then output current) we need a gain of atleast 3-4000 of the output stage (output transistors plus drivers)
Output transistors like MJL2119X are min gain of 25 , which puts a lot of demand on the drivers with respect to gain. especially mentioned MJ1503X.
My feeling is that it is necessary to use a triple darlington (and hence a more complicated design)
This is valid only for a bipolar output of course. For a FET it is a different story (mostly just charging and discharging gate capacitance fast enough
opionions?
Otherwise I am following this with great interest. Seems that LM4702 is a great chip!
As for the LM4702- what happends if one put a current limit transistor which "steals" the base drive , i e outputs of the LM4702 , this transistor driven by voltage over emitter resistors. (the standard configuration).
Will LM4702 output stage tolerate this or will it blow (as one short circuits the outputs via current limit transistor down to earth via the load connected ) , the most likely consequence? Anybody with experience?
Or will we have to resort to more tricky solutions with current limit reistors at the output and carefully designed output stages with high gain, or constant current drives for the chip itself, i e between the chip supply pin and voltage supply or even more exotic ways?
It is a shame that National doesnt mention this in their data sheet. Using a +- 75 V amplifier without current limit and other protection circuits, well I can only say : Stack up on output transistors And probably a few test boards and emitter resistors aswell....
/rickard
It feels like that to utilize the whole voltage range of the amp (and then output current) we need a gain of atleast 3-4000 of the output stage (output transistors plus drivers)
Output transistors like MJL2119X are min gain of 25 , which puts a lot of demand on the drivers with respect to gain. especially mentioned MJ1503X.
My feeling is that it is necessary to use a triple darlington (and hence a more complicated design)
This is valid only for a bipolar output of course. For a FET it is a different story (mostly just charging and discharging gate capacitance fast enough
opionions?
Otherwise I am following this with great interest. Seems that LM4702 is a great chip!
As for the LM4702- what happends if one put a current limit transistor which "steals" the base drive , i e outputs of the LM4702 , this transistor driven by voltage over emitter resistors. (the standard configuration).
Will LM4702 output stage tolerate this or will it blow (as one short circuits the outputs via current limit transistor down to earth via the load connected ) , the most likely consequence? Anybody with experience?
Or will we have to resort to more tricky solutions with current limit reistors at the output and carefully designed output stages with high gain, or constant current drives for the chip itself, i e between the chip supply pin and voltage supply or even more exotic ways?
It is a shame that National doesnt mention this in their data sheet. Using a +- 75 V amplifier without current limit and other protection circuits, well I can only say : Stack up on output transistors
And probably a few test boards and emitter resistors aswell.... /rickard
Finally some more time to 'play' with this design. I will check TOSHIBAs this weekend and let you know how about results.
Rikkitikkitavi, I will probably use optocouplers for current limit and DC protection circuits. There will be small micrcontroller responsible for all the functions. Give me some time and I can also test transistors connected to emiter resistors working as current limiter
I think that DC protection in most important.
Some ON transistors for tests:
Rikkitikkitavi, I will probably use optocouplers for current limit and DC protection circuits. There will be small micrcontroller responsible for all the functions. Give me some time and I can also test transistors connected to emiter resistors working as current limiter
I think that DC protection in most important.
Some ON transistors for tests:
An externally hosted image should be here but it was not working when we last tested it.
Sounds like an interesting protection circuit with optocouplers.
but yes, I agree, DC protection is more important as a few BJT outputs generally are less expensive than most woofers that we use, which of course would meet a violent death at + 70 V DC. Especially 4 ohm.
Question is then what type of protection relay to use at output to make sure that it can cut of without having conntacts welded ...
And a quick relay or fuse and sufficient enough of output transistors a short might, but just might be survivable.
(I woudl expect it though)
now I have holidays for a month, plan to spend much of it hiking in norway so I expect to see some results when back
(part of it will also be spent trying to build a generic amplifier chassi out of 200 mm PVC tube, some 300 VA transformers , a few 15*44*6 cm heatsink and some other stuff I just have got around, i e it will be my first "tube amp"
well, good luck with the construction work!
/rickard
but yes, I agree, DC protection is more important as a few BJT outputs generally are less expensive than most woofers that we use, which of course would meet a violent death at + 70 V DC. Especially 4 ohm.
Question is then what type of protection relay to use at output to make sure that it can cut of without having conntacts welded ...
And a quick relay or fuse and sufficient enough of output transistors a short might, but just might be survivable.
(I woudl expect it though)
now I have holidays for a month, plan to spend much of it hiking in norway so I expect to see some results when back
(part of it will also be spent trying to build a generic amplifier chassi out of 200 mm PVC tube, some 300 VA transformers , a few 15*44*6 cm heatsink and some other stuff I just have got around, i e it will be my first "tube amp"
well, good luck with the construction work!
/rickard
I'm playing with Toshiba MOSFET transistors now - 2SK/2SJ pair. 100mA current per channel (about 3.3V between gates).
This amplifier is really good in this configuration I'm thinking about another simple board like this with some protection circuit.
This amplifier is really good in this configuration
I'm thinking about another simple board like this with some protection circuit. An externally hosted image should be here but it was not working when we last tested it.
Tony said:
I will probably find few LM4780 boards but I don't have any for LM4702. There was only a small quantity of them to make some tests. I would like to make another, similar to this one but with protection circuits and developed supply. This will be a transitory stage between present and my 'monster amp'
I really think that about 120W per channel will be enough for most users. The LM4702 is a stereo amplifier so bridging can give about 200W. Construction is easy and good sounding.
Veteran, That is pretty cool it is working well. Have you tested running in stereo? From youir picture you only have 1 channel built up.
Questions for you (if you will be so kind):
You list on the schematic that T1 and T2 uses BD139, MJE350, sorry if this is a dumb question but which are you using on your board?
Did you do anything to get your board working or was it really this easy to layout and build and it just works? I do not have experience with discretes and am a little nervous about working with the LM4702, like blowing up lots of parts.
Which FETs do you like best?
Thanks and sorry if dumb questions but still just a newbie at discrete stuff.
-SL
Questions for you (if you will be so kind):
You list on the schematic that T1 and T2 uses BD139, MJE350, sorry if this is a dumb question but which are you using on your board?
Did you do anything to get your board working or was it really this easy to layout and build and it just works? I do not have experience with discretes and am a little nervous about working with the LM4702, like blowing up lots of parts.
Which FETs do you like best?
Thanks and sorry if dumb questions but still just a newbie at discrete stuff.
-SL
I'm using BD139 transistors in this design. FETs - 2Sk1530/2SJ201 pair. If you need more power you can connect them in parallel.
Does somebody have some new experience with LM4702? I redesigned first board after my tests:
Now it's more universal.
Does somebody have some new experience with LM4702? I redesigned first board after my tests:
An externally hosted image should be here but it was not working when we last tested it.
Now it's more universal.
Veteran, I have been working with a design that is pretty much the same as your first board. I have tried numberous things to bring the THD down. Based on all I have done I would highly recommend the following:
1. Add a 15 - 47pF cap in parallel with the input resistor that goes to GND. I am using only a 33pF ceramic but it seems to pull the noise down some.
2. I changed the electrolytics on the supply lines of the LM4702 from 100uF to 1uf metalized poly film caps. I also put a 1uF cap of the same type from Vcc - Vee on the LM4702.
3. Add a 1uF film cap across the board terminals right at the terminals.
It looks like you have 2 of the 3 suggestions already on the board. The film caps on the LM4702 supply lines have a big effect, from what I can measure. I am getting THD+N (80kHz) with 1kHz, 8 ohm loads, and both channels on and in phase of less than 0.01% above 1W. It goes as low as 0.008% at higher power. 20kHz distortion is also improved with the above changes from 0.1% down to 0.05%. I think the LM4702 is senstive to layout and so only trial and error will find better performance.
FYI, I am using 10ohm + 0.1uf for snubber values. When I changed to 3.3 ohms I had oscillations. I am also using 220 ohms for gate resistors.
Question for you. I am also using the 1530/201 pair with the Vbe multiplyer as a BD139 part. I have some 1529/200s and 1058/162s to try out. Seems the biggest difference is the gate capacitance which, I would think, would be better if lower. I am still so new to this I am curious what insight you have that I might learn.
If I can find time I want to make my own layout and see how it performs. Always fun to try and make your own. This board is pretty hacked up now so time to make a new one.
-SL
1. Add a 15 - 47pF cap in parallel with the input resistor that goes to GND. I am using only a 33pF ceramic but it seems to pull the noise down some.
2. I changed the electrolytics on the supply lines of the LM4702 from 100uF to 1uf metalized poly film caps. I also put a 1uF cap of the same type from Vcc - Vee on the LM4702.
3. Add a 1uF film cap across the board terminals right at the terminals.
It looks like you have 2 of the 3 suggestions already on the board. The film caps on the LM4702 supply lines have a big effect, from what I can measure. I am getting THD+N (80kHz) with 1kHz, 8 ohm loads, and both channels on and in phase of less than 0.01% above 1W. It goes as low as 0.008% at higher power. 20kHz distortion is also improved with the above changes from 0.1% down to 0.05%. I think the LM4702 is senstive to layout and so only trial and error will find better performance.
FYI, I am using 10ohm + 0.1uf for snubber values. When I changed to 3.3 ohms I had oscillations. I am also using 220 ohms for gate resistors.
Question for you. I am also using the 1530/201 pair with the Vbe multiplyer as a BD139 part. I have some 1529/200s and 1058/162s to try out. Seems the biggest difference is the gate capacitance which, I would think, would be better if lower. I am still so new to this I am curious what insight you have that I might learn.
If I can find time I want to make my own layout and see how it performs. Always fun to try and make your own. This board is pretty hacked up now so time to make a new one.
-SL
Veteran, one other thing, lower gain also helps improve THD measurements. I bumped it down to 28V/V (61.9k/2.2k).
The LM4702 is made for as much power as you want, just parallel up the output stage and set the voltage. I would not try to parallel darlington devices since they have stability issues when running them in parallel. A fully discrete output stage would be fine. FETs may not work well in parallel because of the limited drive current. Get some parts and give it a try.
-SL
The LM4702 is made for as much power as you want, just parallel up the output stage and set the voltage. I would not try to parallel darlington devices since they have stability issues when running them in parallel. A fully discrete output stage would be fine. FETs may not work well in parallel because of the limited drive current. Get some parts and give it a try.
-SL
I forgot to attach schematic diagram for this one. I'm now using 47k/2k resistors to set gain. If somebody needs more power from a single pair - external driver (OPA1632, LM4652) and bridged mode Symmetrical design should sound even better.
You can use 1529/200s. I wanted to try them but they are hard to find here. Input impedance of power stage should be as low as it's possible. Another thing - this power amplifier can work with smaller gain + external preamplifier.
I can send you my board - you will be able to try it and make some test.
One more thing - 1530/201 transistors should be matched
schematic
Symmetrical design should sound even better. You can use 1529/200s. I wanted to try them but they are hard to find here. Input impedance of power stage should be as low as it's possible. Another thing - this power amplifier can work with smaller gain + external preamplifier.
I can send you my board - you will be able to try it and make some test.
One more thing - 1530/201 transistors should be matched
schematic
Help a beginner in discretes again, please. How does one match FET pairs like the 1530/201? I only ordered a few and just stuck them in the board without measuring anything. What does it take to match?
Veteran, where you wanting to send me a board to test? I am not sure who that comment was directed too.
Thanks,
-SL
Veteran, where you wanting to send me a board to test? I am not sure who that comment was directed too.
Thanks,
-SL
SpittinLLama said:
You have to select them for the same RDS-ON.
Yes, I can send you my new board
I will order them after this weekend (there will be a simple PS section). PCB for LM4702 amplifier
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.