😱
Where on earth did you get this info?! 42-45 C recommendation all over my thread.
Lazy,
Sorry that was what I thought was the temp you were recommending. 45C seem more reasonable but what is making you choose that temp, what are the parameters that sets the temperature for best audio?
Sorry that was what I thought was the temp you were recommending. 45C seem more reasonable but what is making you choose that temp, what are the parameters that sets the temperature for best audio?
Are you referring to Junction temperature (Tj), or transistor case temperature (Tc), or to heatsink temperature (Ts)?
Operating temperature ?
The internal silicon will run >50C. You can see this on a properly
monitored computer video card. The actual silicon core will run 20C
hotter than the externals (heat-spreader).
If we are talking about external temperature , this will be the conduction
of the semi's slug to whatever extrusion you have (and it's C/W rating).
PS - all those thermal resistances - die/slug - compound - insulator.
Another factor is your class AB bias. I have 5 pair at 70ma right now.
27c idle - 35c at medium. Never got above 45C (party!!). Thermal
shutdown is set at about 55C.
A single pair MT-200 or TO-264 amp can be made to run at nearly room
temperature with a good enough extrusion.
I've also noticed a BJT amp will run MUCH cooler with >55V rails.
The FET's should not have to run much more than 100ma , should be
just as cool as the BJT.
I suspect Tjmax will be like a video card (>120C) .... 150C max for
a MJL BJT.
OS
Yes , after "digesting" Diptrace , I'll port the chosen IPS's to a interleaved
OPS , same all in one layout as the Honeybadger (but way better).
It will have basic onboard DC shutdown (SS output relays) and can connect to
Vziachenko's "21'st century" protection.
PS - 3 pair MT-200/TO3p/TO264 OPS.
OS
I've got a bunch of models drawn for electrolytic and film caps. The preloaded models leave you cramped for space.
Every little thing has its sweet spot, so has the First One: +/-60 V/280 mA/45 °C
so have humans: 36,5 °C/16-11 kPa
etc.
Sound quality determined by listening tests.
I don't believe the value of these resistors in the OPS are very critical. They provide some decoupling from the raw supply for the driver stages and cap' multiplier.
What is important are the values of R11 and R13 in the input stages.(Symasui example) These are the two resistors that drop the supplies leaving 12 volts across the zeners. See post 294 in the Slewmaster builds thread.
Keith
Jeth , R114/115 values are not critical.
Spooky has the most current draw at 16ma with 2 LTP's and it's servo.
Ohms law says you just get a .16V drop even across a 10R. (.56W dissipation).
Ripple will drop to the predriver/driver with either a larger C111/113 or
R114/115. Even 22R here would be <.4V drop.
Now , after this R/C .... we have the cap multipliers. They "burn" off a
couple volts. R101/102 can be ... 470R/47K for the least 1.6V drop here.
Total , you lose 2V to gain nearly <1mv rail ripple at the IPS .
Going from +/- 35 to +/- 70V .....
All the servo'ed IPS's will need adjustment of the zener shunt regulator
resistors (the 1W one's).
2.7K to 3.3K will establish those 12V supplies at 35V. 3.9K - 4.7K for 60V+.
I usually temporarily tack on a second resistor (to increase I) for my 38V test .
PS - wolverine will work from 20V- 80V rails (no change needed).
Spooky has the most current draw at 16ma with 2 LTP's and it's servo.
Ohms law says you just get a .16V drop even across a 10R. (.56W dissipation).
Ripple will drop to the predriver/driver with either a larger C111/113 or
R114/115. Even 22R here would be <.4V drop.
Now , after this R/C .... we have the cap multipliers. They "burn" off a
couple volts. R101/102 can be ... 470R/47K for the least 1.6V drop here.
Total , you lose 2V to gain nearly <1mv rail ripple at the IPS .
Going from +/- 35 to +/- 70V .....
All the servo'ed IPS's will need adjustment of the zener shunt regulator
resistors (the 1W one's).
2.7K to 3.3K will establish those 12V supplies at 35V. 3.9K - 4.7K for 60V+.
I usually temporarily tack on a second resistor (to increase I) for my 38V test .
PS - wolverine will work from 20V- 80V rails (no change needed).
Keith, Pete, thank you both very much for the clarification. I'm planning on building the Vindicator V(fa) with a 3 pair OPS. Based on your clarifications, I'll go with 6.8E for R114/R115 and 4.7K for R15/R16 (on the Vindicator).
Hari
Hari
Pete, also please confirm that R15/R16 on the Vindicator V(fa) would also be of 1W rating (there's no mention in the schematic that I downloaded).
Thanks
Hari
Thanks
Hari
Ohms law is your friend here , too.
If you have 60V rails , 48V drop through 4.7K = .5W.
Best to derate to a 1W component.
Same for the Zener = >1W .....
OS
If you have 60V rails , 48V drop through 4.7K = .5W.
Best to derate to a 1W component.
Same for the Zener = >1W .....
OS
R=Vin-Vzener/X mA.
Please Pete is the 20mA value the right value for X parameter?
On some of them , dependent on what the rail voltage is.
48V / 4.7K , also equals 10ma/.5W . What the zener shunt is "burning off" (at 60V rails) is about 6ma
This leaves what is actually being used by the IPS/servo. Combined with the
shunt about 12-15ma.
Depending on rail voltage this can change since the shunt will "burn" more
at higher voltages.
5 loads on the IPS cap multipliers -
For example - spookyamp
- 7.4ma VAS (2.1 led's + 5.3ma output)
- 14ma through the 3.9K zener resistors (7ma zener + 3 LTP + 3 op-amp)
I'm using about 22ma per rail total.
Total can change with different VAS/differential current settings.
If you just count the zener I , you might think a smaller resistor will work.
The resistor current actually is the zener,CCS, and op-amp (I) .
In the real world , even a 1W device gets quite warm.
Another "trick" is to just use the highest value R you can to get 12V at
the shunt.
PS - to cover almost anything , a 2W 3.9K will cover 45-75V.
A 1W will do if you calculate the R for the rail voltage.
A 1W even at 3.9K will get hot , but not burn. Mount it above the board.
OS
Last edited:
You say 22mA for the Spooky.
R=Vinp-Vzener/Izener+led+VAS
R=70-12/0.022=2K6 Why you say 3K9?
I missing something?
R=Vinp-Vzener/Izener+led+VAS
R=70-12/0.022=2K6 Why you say 3K9?
I missing something?
You are calculating the 22ma for the zener only.
You forgot Iservo and I -LTP(CCS) .
CSS and the op-amp also need current !!
Cascode LED + the VAS do not use the 12V shunt regulator , but add
to the total IPS I.
(below) with my current 2w 3.9k shunt resistors , I only use 7ma through
the zener.
I've tried 4.7K , then the zener only shunts 1ma. If I adjust my CCS's above
3.5ma with this resistor , voltage will drop to 11.5V. With enough current
draw from the servo + CCS's .... the zener is useless because there is enough
voltage drop through the resistors from the amp itself.
On my present setup , my zener (D5) uses 7.7ma to shunt + 6.4 for the amp/
servo. R23 dissipates @ 14ma. Zener is cold , R23 is very warm (but not hot).
Voltage is rock steady 12.1V. My supplies are +/-73V.
OS
Attachments
OS, in the OPS (and the Vindicator), can I use MJE340/MJE350 in place of KSA1381/KSC3503? These are proving real hard to get.
Thanks
Hari
Thanks
Hari
Actually , it is better to use any >60mhz Ft for the pre-drivers.
340/350 are quite slow <20mhz Ft.
I only use 340/350's for bias generators and cap multipliers.
Here , just Hfe (current gain) matters.
Even MPSA92/42 would be better here. Any
TO-92 or 92L will also work here - not much current , but
watch the pinouts ... (ECB vs.EBC).
You want any >60mhz and >1W device for these pre-drivers.
- 30mhz and 100W for drivers , 30mhz for outputs.
PS - 3 pair OPS's have been run successfully with just a 1837/4793
pair as driver (To-220). BTW , the 1837/4793 toshiba's would be overkill (20w) ...
but also a good pre-driver choice - fast / 100mhz Ft.
Edit- you realize any of the slewmaster VAS's can be biased down to 3-4ma.
At this point , any of the fast TO-126 packages can be replaced with TO-92/L.
Only <1W at 3.5ma .... at 5.5ma+ almost 2W , larger package here.
OS
Last edited: