I was recently inspired by a thread here with a NAD lat-fet conversion, and did some reading on lat-fets/amps, and now I bought two pairs of Exicon ECX10N/P20 lateral fet's to play with, and the idea right now is to put them in a Pioneer amp I have on the shelf:
Thoughts/comments? Any reason this amp is not a good candidate?
Plan is to remove emitter resistors, add larger heat sinks, possibly rework bias spreader, and compensation could be a challenge..
Thoughts/comments? Any reason this amp is not a good candidate?
Plan is to remove emitter resistors, add larger heat sinks, possibly rework bias spreader, and compensation could be a challenge..
Last edited:
No, just want to play around with laterals, and this amp is sitting on a shelf doing nothing. I don't like the sound of the amp, and after some attempts to improve it, I'm still not happy with the sound, even if it got a little better.
The improvement in quality after such operation is highly debatable but the loss in power (~20%) is a certainty.
As you said, the frequency compensation looks strange, I am not sure they really knew what they did at the time. If you really want to plunge into this adventure I strongly recommend a complete spice simulation before heating the soldering iron.
About the Pioneer amp ("I don't like the sound of the amp"):
- odd C348 (C345?) at the emitter of Q304
- C304 "N.M." (not mounted), hf response?
- Q310 & Q312 dumping their collector current into ground through 560Ω resistors...
- Q320 & Q322 are darlingtons?
Lat fets and bipolar amps sound the same to me.
The lat fets would need a higher Vbe multiplier voltage than bipolar.
The lat fets would need a higher Vbe multiplier voltage than bipolar.
H
HAYK
You must first understand the reason of the topology that the designer had in mind. This amp is not a standard Lin topology as the upper and lower branches are referenced to ground. Why?. The designer wanted this amp to low output impedance in open loop. If you consider the Hfe of the output darlington to 1000, the impedance at the bases is 2x22k, this yields 11ohms open loop output impedance. With lateral MOSFETs the gate impedance will determine up to what frequency the open loop output impedance will remain low and resistive maybe 2x100k is enough. This will increase the OLG.
The compensation is standard Miller with 330pf x2 capacitors that can be replaced with 2pole compensation feedbacking the collector to the base of the emitter followers with CRC.
The lateral MOSFETs don't need source resistor and don't need temperature compensation. The colector of the Vbe multiplier can be cut and the bias is done by a fix resistor from gate to gate.
Q308 used as inverter can be replaced with faster 2n5401.
You can also replace the inputs with 2n7000 by shorting the emitter 100ohms resistors.
The compensation is standard Miller with 330pf x2 capacitors that can be replaced with 2pole compensation feedbacking the collector to the base of the emitter followers with CRC.
The lateral MOSFETs don't need source resistor and don't need temperature compensation. The colector of the Vbe multiplier can be cut and the bias is done by a fix resistor from gate to gate.
Q308 used as inverter can be replaced with faster 2n5401.
You can also replace the inputs with 2n7000 by shorting the emitter 100ohms resistors.
Easier to build a real Maplin from scratch. It will outperform any butchering of this Pioneer.
I did consider scrapping the project, and started looking at the Maplin and also looked at other builds, like ESP, and this one was inspiring: https://www.diyaudio.com/community/threads/just-another-lateral-fet-amp.340585/
Anyway, I came back to this conversion, since building an amp from scratch would involve a lot more work and money. My main problem is PCB's, I can't design them, and even if I tried, the result would probably be worse than converting this amp. Several iterations, ordering and waiting etc., so I decided to go ahead and try to convert this one.
I started with a baseline spice simulation, and it looks like it 'could' be stable by just dropping in the lat-fets with PM around 85 and GM abt 8. Loop gain is abt 35dB according to sim.
I'm not looking for ultimate loop gain, I actually found I like the sound of amps with a loop gain of 20-30dB better for some reason, I even like the sound of valve amps.. so it seems distortion is not always an issue for my ears. I found I prefer low output impedance for bass, so I have used AKSA's nested feedback to tune this in several amps intended for full range use. I did it on this amp too, and thought it was an improvement, but not enough. I run an active setup in my main system, so there the amps can be tailored a bit depending on what range and load I use them for.
I played around with the sim a little bit, and found max output swing was around 30V with 40V rails (depending on load), and it could be improved to abt 35V if rails to the earlier stages was increased with 5V. More than that did not seem to improve anything. Anyway, max output is not the aim here, only more pleasing to my ears.
Anyway, I came back to this conversion, since building an amp from scratch would involve a lot more work and money. My main problem is PCB's, I can't design them, and even if I tried, the result would probably be worse than converting this amp. Several iterations, ordering and waiting etc., so I decided to go ahead and try to convert this one.
I started with a baseline spice simulation, and it looks like it 'could' be stable by just dropping in the lat-fets with PM around 85 and GM abt 8. Loop gain is abt 35dB according to sim.
I'm not looking for ultimate loop gain, I actually found I like the sound of amps with a loop gain of 20-30dB better for some reason, I even like the sound of valve amps.. so it seems distortion is not always an issue for my ears. I found I prefer low output impedance for bass, so I have used AKSA's nested feedback to tune this in several amps intended for full range use. I did it on this amp too, and thought it was an improvement, but not enough. I run an active setup in my main system, so there the amps can be tailored a bit depending on what range and load I use them for.
I played around with the sim a little bit, and found max output swing was around 30V with 40V rails (depending on load), and it could be improved to abt 35V if rails to the earlier stages was increased with 5V. More than that did not seem to improve anything. Anyway, max output is not the aim here, only more pleasing to my ears.
Yes but £32 each and that allows you to keep the Lats to do something better in a better chassis with better components 🙂
EBB150 will cost double including shipping and outrageous import taxes thanks to brexit.
The cheapest solution ever would be Maplins with Prasi's layout published here somewhere.
I built those as a sanity project during lockdown. They sound amazingly well for the simplicity.
The cheapest solution ever would be Maplins with Prasi's layout published here somewhere.
I built those as a sanity project during lockdown. They sound amazingly well for the simplicity.
Last edited:
Here is one of my ECX10N20 designs using SPICE.
Very basic amplifier, but thanks to TMC compensation very low distortion = THD 0.00005%
The output stage is simple with one pot setting bias to 200mA.
+/-20V supply which can be increased.
Very basic amplifier, but thanks to TMC compensation very low distortion = THD 0.00005%
The output stage is simple with one pot setting bias to 200mA.
+/-20V supply which can be increased.