Note that simulation shows the DC dissipation of the feedback resistor, with zero output. You have to do some manual calculations to determine the required size of these resistors. Roughly, 20V on 500 Ohms is 0.8 Watts. Besides low noise, low value feedback suffers less from stray capacitance shunting.
Just for the fun of it, I simulated with vertical MOSFETs IRF640 and IRF9640 and surprisingly, the outcome is better than with the Exicons. THD is lower and the harmonics are perfectly falling.
OK, it is only simulation, but given the low price of them, it might be worth a shot.
OK, it is only simulation, but given the low price of them, it might be worth a shot.
When you are exploring VAS currents, be aware of the impact the degeneration resistors have (R9/10 on your schematic). As VAS currents change, these should be assessed. Ideally the 10 x Re heuristic would apply, but I found that DC offset and bias where too unstable with regard to temperature changes. I tried to find a balancing point where thermal stability became satisfactory, but performance didn't suffer too much.
Device gain impacts this too. TTC004B/TTA004B have higher gain that SC3503/SA1381. So these will cause more temp change for the same degeneration.
I see thermal stability as the change from cold to hot states when the amp powers on, and changes in ambient temp.
To model this in LTSpice, you have to provide temperature for all the transistors in the design. I use parameter in Spice that combine ambient temp with the estimated junction temperature rise. And for the Junction temp rise, I have a scaling factor I can use to disable this to simulate starting cold.
To get the temp rise, you have to make some educated guesses on how high the junction temps will rise based on the heatsinks you will be using. The outputs will also change with the load placed on them. Not a perfect science.
Device gain impacts this too. TTC004B/TTA004B have higher gain that SC3503/SA1381. So these will cause more temp change for the same degeneration.
I see thermal stability as the change from cold to hot states when the amp powers on, and changes in ambient temp.
To model this in LTSpice, you have to provide temperature for all the transistors in the design. I use parameter in Spice that combine ambient temp with the estimated junction temperature rise. And for the Junction temp rise, I have a scaling factor I can use to disable this to simulate starting cold.
To get the temp rise, you have to make some educated guesses on how high the junction temps will rise based on the heatsinks you will be using. The outputs will also change with the load placed on them. Not a perfect science.
Looked at the dissipation for the feedback resistor at 500R running a sine wave at 40W. In actual use, it should never see anything this extreme.
Dissipation is 440mW RMS or 875mW Peak.
Drop output to 1/3 full power (around 15W) and dissipation is 170mW RMS / 330mW Peak.
I'm comfortable with the recommended 1W rating.
Now if you increase the resistors to 5k / 1k you can easily get away with a typical 1/4W resistor. And smaller feedback cap.
Dissipation is 440mW RMS or 875mW Peak.
Drop output to 1/3 full power (around 15W) and dissipation is 170mW RMS / 330mW Peak.
I'm comfortable with the recommended 1W rating.
Now if you increase the resistors to 5k / 1k you can easily get away with a typical 1/4W resistor. And smaller feedback cap.
That would be an interesting fork to this design. However, the bias network would need to change. Laterals have a positive Tempco above about around 120mA, so they can get away with a simple resistor to set the bias. Verticals would need a different approach due to their negative tempco.
[For the lateral MOSFET gate stoppers, there isn't a clear consensus among designers on exact values. I've seen anything from 100R to 470R in these positions. I suggest 330/220 for TO-247 devices and you are suggesting 330/270.] you are slowing the slower one down more by having bigger gains topper on it. Swap them round and you compensate for their charger speed imbalance. 270R at top and 330pR at bottom.
{The MOSFET gate-source and drain-source caps are unnecessary if Exicon outputs are used. Not sure if this is the case with other out of production devices - assuming a builder was using NOS parts.} All Lateral MOSFETS follow same convention unless fake. BTW NOS wont work in a Lateral MOSFET design, you may as well use a transistor because you have to change everything.
[D7/8 protect the rails if there is a fault on one rail. Bob Cordell suggests their use. Just protection - not needed for the amp to function normally. Many designs do not use these.} That is exactly the reason for the body diode built into the FET, also called free wheeling diode in cases of excessive inductor loads. I don't think Bob Cordel is aware of this fact exists in Lateral MOSFETS, others such as HEXFETS, etc does not have it and need be added, that is what Bob is used to, Same with the base protection diodes at 14V. Lateral Mosfets has them built in. You can hardly destroy a Lateral MOSFET. It functions as a variable resistor not a semi conductor. Save the components and board space, more parts is not better it is superfluous.
The optimum bias current (Is/Id) is 100mA according to studies done by the manufacturer, therefor VR3 can be substituted wit a 1n$!$^and it would be at optimal. Should you want it to be higher biased in class A, use two diodes in series but they will run hotter. 1 diode is optimum.
We never used D9 or 10, even when at Hewlett Packard and also very sensitive military designs, I cannot see its function. Please explain exactly it does, in mains operated equipment. The only reason for those diodes is to protect the amplifier against power potential reversal in battery operated stuff when the battery is reversed by accident on installation and the main fuses will blow.
Never thr less you are free to add and take away from the design as you please. I will build it in accordance with my experience. I in any case like a PCB laid out with both channels mirrored and can mount the whole amp on a single heat sink and bury it inside the box. These hot rod amps with fins on the sides never appealed to me.
Never the less nice job.
{The MOSFET gate-source and drain-source caps are unnecessary if Exicon outputs are used. Not sure if this is the case with other out of production devices - assuming a builder was using NOS parts.} All Lateral MOSFETS follow same convention unless fake. BTW NOS wont work in a Lateral MOSFET design, you may as well use a transistor because you have to change everything.
[D7/8 protect the rails if there is a fault on one rail. Bob Cordell suggests their use. Just protection - not needed for the amp to function normally. Many designs do not use these.} That is exactly the reason for the body diode built into the FET, also called free wheeling diode in cases of excessive inductor loads. I don't think Bob Cordel is aware of this fact exists in Lateral MOSFETS, others such as HEXFETS, etc does not have it and need be added, that is what Bob is used to, Same with the base protection diodes at 14V. Lateral Mosfets has them built in. You can hardly destroy a Lateral MOSFET. It functions as a variable resistor not a semi conductor. Save the components and board space, more parts is not better it is superfluous.
The optimum bias current (Is/Id) is 100mA according to studies done by the manufacturer, therefor VR3 can be substituted wit a 1n$!$^and it would be at optimal. Should you want it to be higher biased in class A, use two diodes in series but they will run hotter. 1 diode is optimum.
We never used D9 or 10, even when at Hewlett Packard and also very sensitive military designs, I cannot see its function. Please explain exactly it does, in mains operated equipment. The only reason for those diodes is to protect the amplifier against power potential reversal in battery operated stuff when the battery is reversed by accident on installation and the main fuses will blow.
Never thr less you are free to add and take away from the design as you please. I will build it in accordance with my experience. I in any case like a PCB laid out with both channels mirrored and can mount the whole amp on a single heat sink and bury it inside the box. These hot rod amps with fins on the sides never appealed to me.
Never the less nice job.
Last edited:
Yes
Ignore my comment on D9 and 10 it may prevent a DIYer to accidentally reversing the power supply and blow his amp, The fuses will blow instead. I only thought of battery reversal and not a stupid mistake by the DIYer. (Although the body diode will also protect against accidental power reversal)
Ignore my comment on D9 and 10 it may prevent a DIYer to accidentally reversing the power supply and blow his amp, The fuses will blow instead. I only thought of battery reversal and not a stupid mistake by the DIYer. (Although the body diode will also protect against accidental power reversal)
Last edited:
Hi!
I have a silly question, I know, so please be nice...
What would I have to change to be able to use SJ73 + SK146 for inputs?
To answer the "why": because I have quite a few of those.....
Is a change in R6 pot enough?
Thanks a lot!
I have a silly question, I know, so please be nice...
What would I have to change to be able to use SJ73 + SK146 for inputs?
To answer the "why": because I have quite a few of those.....
Is a change in R6 pot enough?
Thanks a lot!
I will run 25V rails, I am not a power freak, my room is the size of a pee pot. I could run 48V by cascoding the input. I also have hundreds of these things.
Morten, did you follow Brians question or have you gone to sleep. It is 00:35 here and I am in my 70s
I don't know if this was an answer to my question, but it is crucial. I forgot the 25V spec. Will +/- 24V be ok?
Morten, If you run lower voltage you my bias high into class A that is a BIG +plus. The LatMOSFETS sounds stunning in classA
EDIT: Because they operate as variable resistors they have an incredibly smooth soft sound, I hate to say this "valve like"I know this is BS but you get my point.
EDIT2: That is exactly what I will do run at 24V rails so nothing radical to change,
EDIT: Because they operate as variable resistors they have an incredibly smooth soft sound, I hate to say this "valve like"I know this is BS but you get my point.
EDIT2: That is exactly what I will do run at 24V rails so nothing radical to change,
How low do I really have to go, for safety? 20V? Less?
I already have a few class A amps, so was thinking something cooler.... and less bulky!
I already have a few class A amps, so was thinking something cooler.... and less bulky!
Morton, don't be a whimp now. The sun is probably still shining up in the north
EDIT: 24 should be fine.
EDIT2: Then Cascode will be the answer, but the PCB will need modification
EDIT: 24 should be fine.
EDIT2: Then Cascode will be the answer, but the PCB will need modification