Here are just my personal thoughts, even though this is not my undertaking.
In the end the reality will come out as it should.
Given 1000 FETs matched under operating conditions, this is a one-time exercise that I doubt anyone else would repeat soon.
Firstly all FETS of the same type would be sorted to quads of say within +/-10mV. The rest are rejects.
(Why 10mV, because in my experience it is nearly impossible to get better repeatability, given so many variables)
Then there will be x% of NNNNPPPP matches to say within 40mV, y% to say 80mV, z% to ..., etc.
And then it will only be logical that the prices are proportional to 1/x, 1/y, ...... etc.
The total sum should in the end pay for the material, transport, a bit of labour, rejects, etc., so that Nic the Lion Heart does not suffer a loss.
And I am sure depending on the rarity of perfect NNNNPPPP matches, the price will determined the demand.
If not, then it will likely be first come first serve, with the Case GB members given priority.
Right, Nic ?
Patrick
PS what numbers would x, y, z, ...be, we shall know when it is done, which won't be anytime soon.
That also means it is impossible to put a price tag until then.
In the end the reality will come out as it should.
Given 1000 FETs matched under operating conditions, this is a one-time exercise that I doubt anyone else would repeat soon.
Firstly all FETS of the same type would be sorted to quads of say within +/-10mV. The rest are rejects.
(Why 10mV, because in my experience it is nearly impossible to get better repeatability, given so many variables)
Then there will be x% of NNNNPPPP matches to say within 40mV, y% to say 80mV, z% to ..., etc.
And then it will only be logical that the prices are proportional to 1/x, 1/y, ...... etc.
The total sum should in the end pay for the material, transport, a bit of labour, rejects, etc., so that Nic the Lion Heart does not suffer a loss.
And I am sure depending on the rarity of perfect NNNNPPPP matches, the price will determined the demand.
If not, then it will likely be first come first serve, with the Case GB members given priority.
Right, Nic ?
Patrick
PS what numbers would x, y, z, ...be, we shall know when it is done, which won't be anytime soon.
That also means it is impossible to put a price tag until then.
Last edited:
The Hi-Tech, hi-cost solution :
Temperature controlled heater block using a PID controller.
The low-tech, near zero cost solution :
A very large aluminium block (say 10kg min) in your dosmetic oven preheated and soaked at test temperature for say 2 hours.
Assuming of course your Mrs will let you use the oven to bake FETs.
(No problem for me.)
And make sure your CT (whatever brand) does not drift all over in continuous operation, especially anything to do with current sensing.
I guess you also know how to calculate dissipation of the CT power transistors, and what cooling they would then need.
Patrick
Temperature controlled heater block using a PID controller.
The low-tech, near zero cost solution :
A very large aluminium block (say 10kg min) in your dosmetic oven preheated and soaked at test temperature for say 2 hours.
Assuming of course your Mrs will let you use the oven to bake FETs.
(No problem for me.)
And make sure your CT (whatever brand) does not drift all over in continuous operation, especially anything to do with current sensing.
I guess you also know how to calculate dissipation of the CT power transistors, and what cooling they would then need.
Patrick
Last edited:
Nope. I'm playing with the Locky_z CT and I'm having fun and learning, but the plan is to use something more serious for these mosfets.
Just as an exercise, take an example of a 11W ceramic wire wound power resistor.
At half power (5.5W), its temperature rise is something like 150°C (in free air, unobstructed).
With a tempco of say 400 ppm/K, that would be a dR of 6%.
If the resistor is used to measure a nominally 2A current, then the error would be 0.12A.
To match to say 10mV Vgs, your measurement uncertainty should be at least half that, i.e. 5mV.
The transconductance of 2SK1530 at 2A is about 2.5S.
So 5mV in Vgs is equivalent to 12.5mA, or 0.0125A (0.625% of 2A).
Yes, there are ceramic wire wound resistors with zero tempco.
Maybe the one in your CT is. I hope so.
And you can of course also keep the temperature low by using the CT only in pulse mode.
Then the resistor never gets heated up in the first place.
BTW, to measure the 2SK1530 in the full operating range of the F5X, you should measure 0~4A Id.
If your current sensing resistor is say 5R, the mean dissipation is something like 27W.
To keep dT to 150°C, you would need a 55W resistor.
Patrick
At half power (5.5W), its temperature rise is something like 150°C (in free air, unobstructed).
With a tempco of say 400 ppm/K, that would be a dR of 6%.
If the resistor is used to measure a nominally 2A current, then the error would be 0.12A.
To match to say 10mV Vgs, your measurement uncertainty should be at least half that, i.e. 5mV.
The transconductance of 2SK1530 at 2A is about 2.5S.
So 5mV in Vgs is equivalent to 12.5mA, or 0.0125A (0.625% of 2A).
Yes, there are ceramic wire wound resistors with zero tempco.
Maybe the one in your CT is. I hope so.
And you can of course also keep the temperature low by using the CT only in pulse mode.
Then the resistor never gets heated up in the first place.
BTW, to measure the 2SK1530 in the full operating range of the F5X, you should measure 0~4A Id.
If your current sensing resistor is say 5R, the mean dissipation is something like 27W.
To keep dT to 150°C, you would need a 55W resistor.
Patrick
Last edited:
Sorry for the mistake NicMac. Your setup sounds interesting, any details.
Thanks for the example Euvl. I will be going through the curve tracer thread to get as much info as possible. I know others have posted about improvement possibilities. I intend to use it to match for my GB and from what others have told me, it has performed very well. Sorry for OT post.
Thanks for the example Euvl. I will be going through the curve tracer thread to get as much info as possible. I know others have posted about improvement possibilities. I intend to use it to match for my GB and from what others have told me, it has performed very well. Sorry for OT post.
I thought we were going for 3point matching?
If that is still the case then I would be happy with 1A, 2A & 3A as the 3 points. That covers the majority of all the audio output from the F5x.
That range gives + or - 2Apk to the speaker.
If the 3point match is good @ those 3 set points then the matching is still going to be acceptable at 0.5A and 3.5A. That takes the F5x output to + or - 3Apk.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.