[scootsit]

Quote:

Scootsit. I'm not sure that's going to be any use in power applications, it looks to me that its not a discrete device so much as another way of packing more transistors into a memory chip.

I understand that, this was more theoretically speaking than in practice.
 
I realize I was unclear in my statement.

 

[DingoSmuggler]

Quote:

Hey guys, have you heard that Intel recently developed a new MOSFET? The transistor is evidently the same as the one in the ivy bridge chips. It allows for 37% faster switching and 50% less power use. They're called tri-gate transistors. Any chance one could be popped into a Millett? More here: http://en.wikipedia.org/wiki/Multigate_device#Tri-gate_transistors
 
Also, if they use that little power, wouldn't this be a great transistor for portable applications?
(If the transistors themselves were commercially available, of course)
 
 
Sorry for the way-off-topic-post

digital circuits are concerned with getting a transistor fully on and fully off as quickly and efficiently as possible, linear amp circuits are concerned with the area in-between.
A good switching mosfet doesn't mean a good linear mosfet.

 

[tomb]

Quote:

digital circuits are concerned with getting a transistor fully on and fully off as quickly and efficiently as possible, linear amp circuits are concerned with the area in-between.
A good switching mosfet doesn't mean a good linear mosfet.

Yeah, I don't have nearly the knowledge that you guys do about MOSFETs, but I was going to say ... everything we do is to keep the dang things from "switching" in the first place.

 
Still, I'm glad scootsit is keeping watch and asking interesting questions!
 
Here's one that I have that I should've asked AMB but didn't:
Why are the MOSFET complements in the M3 and B22 (and the MOSFET_MAX

) the Z24 and 9Z34 .. instead of the Z24 and 9Z24?

 

[tdubl07]

Finally got my mini max finished last night. But not before I realised I soldered the transistors on the wrong side of the heatsinks. Apparently the only part of the instructions i missed. That was fun unsoldering those :eek: Sounds great though now that it's done

 

[KimLaroux]

Quote:

...
Here's one that I have that I should've asked AMB but didn't:
Why are the MOSFET complements in the M3 and B22 (and the MOSFET_MAX

) the Z24 and 9Z34 .. instead of the Z24 and 9Z24?

 
I'd guess it has to do with those specs:.
 

​	IRFZ24​	IRF9Z24​	IRF9Z34​
RDSon (ohms)​	0.07​	0.28​	0.14​
Transconducatance​	4.5​	1.4​	5.9​

 
Which shows the 9Z34 matches better to the Z24 than the 9Z24 does.
 
Though when you compare reaction times and pins capacitance, the 9Z34 is not a better match. But I think those specs can be compensated for in the circuit, while the ones in the table cannot.

 

[tomb]

Quote:

 
I'd guess it has to do with those specs:.
 

​	IRFZ24​	IRF9Z24​	IRF9Z34​
RDSon (ohms)​	0.07​	0.28​	0.14​
Transconducatance​	4.5​	1.4​	5.9​

 
Which shows the 9Z34 matches better to the Z24 than the 9Z24 does.
 
Though when you compare reaction times and pins capacitance, the 9Z34 is not a better match. But I think those specs can be compensated for in the circuit, while the ones in the table cannot.

That's as good an explanation as any.  Yet, I'm not so sure that even BJT complements are that close to each other.  I would think reaction times are the most important.
 
The capacitance differences are definitely compensated in the circuit.  We do the same thing in the MOSFET-MAX:

A 220pF capictor is bridged across the Z24's.  Again, one would think that compensating in that way - or especially compensating for resistance/transconductance differences would be even easier ... but not reaction times.

 

[tdubl07]

Hi Guys, question on biasing the tubes on the mini max. If I set the tube bias after the amp has been on and listening to it for about 30 min I set the bias to 13.5 on each tube. If I turn it off for a while and then back on, the biases are up to 14.5-14.8 and then slowly settle unitl its warmed up. Is this correct procedure, or do you bias them when cold right after turning on the amp? Sorry if this is a dumb question but I'm new to the tube world

 

[KimLaroux]

Well reaction times are measured as the time it takes for the MOSFET to switch between 10% and 90% of the difference between Vgs and Vds:
 
 

 
If you consider the times (td, tf) as independent, fixed values, then they create a slope that is dependent on the difference between Vds and Vgs. The higher the difference, the steeper the slope is. Since audio applications, especially headphones ones, don't need the full switching from 10% to 90%, you can cheat by raising the voltages and only using a part of the slope. This way you can match the slopes of different parts, even if their rise and fall times aren't the same.
 
But then I'm unsure if the times are independent on the difference between the gate and the drain. The datasheets suggests so, anyways.
 
In reality though, NFB should take care of all of this...
 
Thinking about it... switching times are probably just not a concern here. We're talking about differences in the nanoseconds. Just to put that in perspective, one nanosecond is the cycle time of a 1 Ghz waveform... 

 

[tomb]

Quote:

Well reaction times are measured as the time it takes for the MOSFET to switch between 10% and 90% of the difference between Vgs and Vds:
 
 

 
If you consider the times (td, tf) as independent, fixed values, then they create a slope that is dependent on the difference between Vds and Vgs. The higher the difference, the steeper the slope is. Since audio applications, especially headphones ones, don't need the full switching from 10% to 90%, you can cheat by raising the voltages and only using a part of the slope. This way you can match the slopes of different parts, even if their rise and fall times aren't the same.
 
But then I'm unsure if the times are independent on the difference between the gate and the drain. The datasheets suggests so, anyways.
 
In reality though, NFB should take care of all of this...
 
Thinking about it... switching times are probably just not a concern here. We're talking about differences in the nanoseconds. Just to put that in perspective, one nanosecond is the cycle time of a 1 Ghz waveform... 

Well, I think the Class A bias would take care of that, too - switching should be removed from the scenario altogether.

 

[lmisner]

Is ther a downside to bypassing RB14L/R or reducing it to 1 or 2R?
 

 

[tomb]

Quote:

Is ther a downside to bypassing RB14L/R or reducing it to 1 or 2R?
 

If you're not using MOSFETs, then no - I've often used nothing but jumpers at RB14.  However, I'm not sure the MOSFET-MAX is stable without about 22R on the output.  Now, I would encourage you to experiment if you're building a MOSFET-MAX.  It's pretty easy to confirm by listening for less bass - that's an indication the MOSFETs are oscillating.  It may be that a higher resistance is needed for RB10 and RB11 - perhaps a 3.3R or 4.7R might make the MOSFETs stable at their output, instead of hanging 22R resistors on the outside.

 

[lmisner]

Ok, thanks.  I have the mosfet version. Never did like the 22R, but figured it was there for protection or phase margin.  Guess I'll hook my scope up and see what I get when I lower the value.

 

[thomaskuhn]

I am surprised that this amp still has a very active forum going for it!  Well, I am finally having to fix something in my MAX that I have had since the first boards were printed.  I have used it ever since in my office.  Well, one thing that always drove me crazy was that I put the power switch "upside down." It always feels more natural to push in the top of a rocker switch to power on.  So I decided to, "ah, 1,2,3" Ill just undo the thumb screws, and flip it.  Well...  The power socket broke apart into about 4 pieces.  I looked on the BOM to get the part number, but notice that and updated BOM had a radio shack all metal socket that could be used. It was easier fro me to go to RS, vs ordering.   My concern is that the original  BOM Socket shows a very clear note... "Must be isolated from Case", but the RS part is all metal.
 
I already have the RS part, so should I isolate it from the case, or will it work fine the way it is?
 
Thanks,
 
Tom

 

[tomb]

  I am surprised that this amp still has a very active forum going for it!  Well, I am finally having to fix something in my MAX that I have had since the first boards were printed.  I have used it ever since in my office.  Well, one thing that always drove me crazy was that I put the power switch "upside down." It always feels more natural to push in the top of a rocker switch to power on.  So I decided to, "ah, 1,2,3" Ill just undo the thumb screws, and flip it.  Well...  The power socket broke apart into about 4 pieces.  I looked on the BOM to get the part number, but notice that and updated BOM had a radio shack all metal socket that could be used. It was easier fro me to go to RS, vs ordering.   My concern is that the original  BOM Socket shows a very clear note... "Must be isolated from Case", but the RS part is all metal.
 
I already have the RS part, so should I isolate it from the case, or will it work fine the way it is?
 
Thanks,
 
Tom

 
You definitely need to isolate the power socket if it is metal.  The Millett MAX uses an AC power supply.  If it's not isolated from the case, your amp's ground will be infected with 60 Hz, possibly resulting in hum.

 

[thomaskuhn]

AH, that's right.  Now I remember the AC power supply.  Thanks!!