That's the TA-3600 KIT.
It's like Phase Linear 700/II.
On the margin: TIP29C/30C and BD245C/250C - these are only 100V Vceo devices!
And hey, in the input 470p and 10k gives only 34kHz.
In the feedback: 270 and 33u, that's 18Hz... Not much...
This amp's bandwith is a tragedy, i think!
And 2 pair of MJ15003 isn't much for 350W/4ohm, at 45° or 60° phase angle. See SOA!
It's like Phase Linear 700/II.
On the margin: TIP29C/30C and BD245C/250C - these are only 100V Vceo devices!
And hey, in the input 470p and 10k gives only 34kHz.
In the feedback: 270 and 33u, that's 18Hz... Not much...
This amp's bandwith is a tragedy, i think!
And 2 pair of MJ15003 isn't much for 350W/4ohm, at 45° or 60° phase angle. See SOA!
"Please note that with regard to SOA, second breakdown and phase angle you need at least 6 pairs of output devices (Vce 250V) to avoid danger "
True, but with a good dual-slope limiter there are many good PA amps with three pair of MJ15024/25 running off of ±75V.
Believe in both a belt and suspenders?
I built up a ±80V amp with 10 pair of MJ15024/24 for outputs, another pair as drivers, and the 2N3584/2N6241 as predrivers.
True, but with a good dual-slope limiter there are many good PA amps with three pair of MJ15024/25 running off of ±75V.
Believe in both a belt and suspenders?
I built up a ±80V amp with 10 pair of MJ15024/24 for outputs, another pair as drivers, and the 2N3584/2N6241 as predrivers.
djk said:
Ah the joys of exceptional cases...
.I have used a similar circuit with 70vce rails. It gave 350 watts into 4 ohms and 240 into 8 ohms. Then I added 2 more pairs of Motorola 15003/4's and the suckers kept alive as I jacked up the rails...higher and higher I went.
In the end I had them at +/- 90vdc rails. 4 pairs per channel. Cranked out 700 watts into 4 ohms. No worries.
Caveat Emptor. Initially I had non MOT devices which burnt out as my vendor ran out of MOT's at higher Vce's However when I switched all to MOT, they were fine.
I know what their Vce rating is and I know they ran fine into +/-90vdc.
So this circuit is not all that outlandish. You guys underrate the MOT MJ15003/4's
K-amps said:
So this circuit is not all that outlandish. You guys underrate the MOT MJ15003/4's
IMHO if you want reliability at driving larger phase angle impeadence, it is better to underate in general, also taking into account thermal derating...which is better with TO-3 case but still to be taken into figure if heatsinking is only moderate.
The
Max VCE value in short data is set to a value
that has got a reasonable safety margin.
In a mass production there will be a few transistors that wont take more than this voltage.
These will determine the value that manufacturer can promise, in worst case.
But most exemplars will take more
and some few in a mass production line will take Considerably MORE voltage.
Even for a longer time, hours, continuous output at maximal levels.
-----------------------------
- Now, different manufacturers has got different qualities in production methods.
There are no-names manufacturers where quality control is not as extensive.
The difference between best and worst exemplars will be a wide spread.
- There are others, which put more effort on high average quality.
To those we can count well known and respected brands like Motorola.
There may even be some, that does not sell those exemplars with lower performance.
They just sort them out, before delivery to public distributors.
Those 'bad exemplars' will be delivered under no-names label,
for low price bargain resellers.
In more detailed databooks, you see often two values for max current.
For small signal transistors.
For example BC547 has got:
IC = 100 mA, max average collector current.
ICM = 200mA, max peak shorter spikes collector current.
Bottomlines:
- Short form data for semiconductors are not absolute values.
- No exemplar will match every single data.
- Near all exemplars will match within the MIN MAX values.
- Most exemplars will have the TYPE value.
- Many exemplars will have considerably better performance than worst case value.
lineup
Max VCE value in short data is set to a value
that has got a reasonable safety margin.
In a mass production there will be a few transistors that wont take more than this voltage.
These will determine the value that manufacturer can promise, in worst case.
But most exemplars will take more
and some few in a mass production line will take Considerably MORE voltage.
Even for a longer time, hours, continuous output at maximal levels.
-----------------------------
- Now, different manufacturers has got different qualities in production methods.
There are no-names manufacturers where quality control is not as extensive.
The difference between best and worst exemplars will be a wide spread.
- There are others, which put more effort on high average quality.
To those we can count well known and respected brands like Motorola.
There may even be some, that does not sell those exemplars with lower performance.
They just sort them out, before delivery to public distributors.
Those 'bad exemplars' will be delivered under no-names label,
for low price bargain resellers.
In more detailed databooks, you see often two values for max current.
For small signal transistors.
For example BC547 has got:
IC = 100 mA, max average collector current.
ICM = 200mA, max peak shorter spikes collector current.
Bottomlines:
- Short form data for semiconductors are not absolute values.
- No exemplar will match every single data.
- Near all exemplars will match within the MIN MAX values.
- Most exemplars will have the TYPE value.
- Many exemplars will have considerably better performance than worst case value.
lineup
Selection is always an option for those who have the equipment and a sufficient stock to work from. The average home constructor or small repair shop generally doesn't.
Keep in mind most transistor testers measure Vceo with a narrow pulse (say 300µS) and a limited duty cycle (say 2%), and not forward biased (like an audio amp) , and that secondary breakdown may remain unchanged in samples that measure higher than normal Vceo.
In the past I have had problems at ±57V with a Motorola part with a 120V Vceo and 180W SOA at 60V rating. In theory this should have been within the capabilitues of the device. In practice it was best to run no more than ±52V without selection of the NPN devices, the PNP were always fine.
Keep in mind most transistor testers measure Vceo with a narrow pulse (say 300µS) and a limited duty cycle (say 2%), and not forward biased (like an audio amp) , and that secondary breakdown may remain unchanged in samples that measure higher than normal Vceo.
In the past I have had problems at ±57V with a Motorola part with a 120V Vceo and 180W SOA at 60V rating. In theory this should have been within the capabilitues of the device. In practice it was best to run no more than ±52V without selection of the NPN devices, the PNP were always fine.
djk said:
Hey.
Even if there is some MAX Data like gain, Vceo, Ft .... that can be override, exceeded by some 'good exemplars'
I have a feeling there are some MAX Data that NOT should be challenged.
Such data, I think, is power data = thermal characteristics.
I think these is very constant values, and most every transistor will have the same.
Example of such important data, in MJ21194
Such values we should NOT override!
Much better to construct with a good safety margin for these values.
Max inner, junction, temperature in MJ21194 is +200 °C.
But when determine heatsinks needed, we should not have have any higher
worst case target value, than +100 °C (maybe +125 °C as this is a TO-3, +200 °C unit)
Most other CASES like TO247, TO220 etc
have maximal junction temperature 50 degrees lower = +150 °C
For Cooling needed, we can have target values of max junction temperature = +80 °C
(maybe +100, +110 if transistor is not used in Amplifier output stage, but some other circuit).
The transistor will work less good and so cause more distortion if running very hot.
Even if temperature is well below max +150 °C.
------------------------------
As a final note:
Nelson Pass has told he runs his Class A HEXFET at a comparatively very low junction temperature.
He use very good cooling, big heatsinks and he is soon to parallel his output IRFXXX devices.
This way his output transistors, even if running a lot of Class A power,
will work at a very moderate temperature and so keep the good performance.
lineup
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.