Electronics > Beginners
PS limiting current for no reason?
<< < (3/3)
spec:

--- Quote from: Sylvi on November 08, 2018, 05:50:41 am ---What is the love for 2N3055 everyone has? They are not great devices and lose gain at currents much below their rating. Why not use modern devices with sustained-beta?

--- End quote ---

They are dirt cheap, rugged, have a reasonably low thermal resistance, junction to case, and they have a maximum junction temperature of 200 deg C. They also have a TO3 case which is good for heatsinking.

What more could you ask for? Well, a better frequency response, more consistency from maker to maker and batch to batch, lower saturation and lower VBE. And, as you say a flatter HFE curve with current changes.

I feel the same way as you about the 2N3055, but they are not too bad for this application. ;D
IanMacdonald:
BDY57/58 are fantastic power devices, especially for amplifiers. They cost a fair bit though.  Modern devices, especially FETs, suffer from some dreadful SOA (secondary breakdown) limits. Another good and cheap muscle BJT is the 2N3773. This is used in many guitar amps. 

As for the 723 problems, if you check the IC schematic you'll see that your arrangement puts a quadruple darlington between the supply and output. That's gonna lose a fair bit of voltage. It may be just that this burden voltage is too great. If you replaced your driver transistor with a PNP, and fed its base from pin 7 (see the application notes fig 20) that would drastically reduce the burden voltage.  As mentioned you should have a turn-off resistor between base and emitter of the driver to stop stray currents or capacitance from turning it on. Say 470R. (Not critical)

http://www.ti.com/lit/ds/symlink/lm723.pdf

Other point, a crowbar which shorts the output can be dangerous if the PSU is used to charge a battery and there is no fuse after this point.
spec:

--- Quote from: IanMacdonald on November 08, 2018, 05:43:35 pm ---BDY57/58 are fantastic power devices, especially for amplifiers.
--- End quote ---
I remember those :) 


--- Quote from: IanMacdonald on November 08, 2018, 05:43:35 pm ---Modern devices, especially FETs, suffer from some dreadful SOA (secondary breakdown) limits.
--- End quote ---
MOSFETs also suffer from huge parasitic capacitances, sometimes in the order of nano farads, which cause all kinds of problems


--- Quote from: IanMacdonald on November 08, 2018, 05:43:35 pm ---Another good and cheap muscle BJT is the 2N3773. This is used in many guitar amps. 
--- End quote ---
Know them well. :)


--- Quote from: IanMacdonald on November 08, 2018, 05:43:35 pm ---As for the 723 problems, if you check the IC schematic you'll see that your arrangement puts a quadruple darlington between the supply and output. That's gonna lose a fair bit of voltage. It may be just that this burden voltage is too great.
--- End quote ---
That is what my fag packet voltage budget calcs show, and what I was hinting at. The 723, itself has a 3V dropout.  ::)


--- Quote from: IanMacdonald on November 08, 2018, 05:43:35 pm ---If you replaced your driver transistor with a PNP, and fed its base from pin 7 (see the application notes fig 20) that would drastically reduce the burden voltage.
--- End quote ---
Yes, that would improve the DC conditions, but it also tends to increase the open loop gain and can lead to frequency stability problems, especially when the constant current shunt transistor kicks in. Many LDO regulators use PNP transistors to get a low voltage overhead with the result that they are a touch critical. 

Back on to power BJTs, here are my favorites:

ON MJL3281A/MJL1302A (TO264) These are the blue blood of low distortion output transistors- notice the flat HFE and high frequency response. It is not unusual to greatly improve an audo amp just by fitting these beauties.

ON MJ15003G/MJ15004G (TO3) similar idea, with a bit more clout.

ON MJL4281/MJL4282 (TO247) more clout.

Must stop now. :palm:

 
IanMacdonald:
"If you replaced your driver transistor with a PNP, and fed its base from pin 7 (see the application notes fig 20) that would drastically reduce the burden voltage.

Yes, that would improve the DC conditions, but it also tends to increase the open loop gain and can lead to frequency stability problems, especially when the constant current shunt transistor kicks in. Many LDO regulators use PNP transistors to get a low voltage overhead with the result that they are a touch critical. "

Indeed, when you add more loop gain and delay to any opamp arrangement you may need to increase the compensation to keep things stable. That is true anyway though. With a driver and output stage both using relatively slow transistors, the standard value of compensation cap will likely be a tad too small. Main thing is not to increase it beyond the point where overshoots start to show when the load is removed.
spec:

--- Quote from: IanMacdonald on November 08, 2018, 11:25:31 pm ---Indeed, when you add more loop gain and delay to any opamp arrangement you may need to increase the compensation to keep things stable. That is true anyway though. With a driver and output stage both using relatively slow transistors, the standard value of compensation cap will likely be a tad too small. Main thing is not to increase it beyond the point where overshoots start to show when the load is removed.

--- End quote ---
You obviously have some experience with compensating PSUs. The problem is that with lab type PSUs you never know what will be connected to the output  terminals: it could be anything from light restive loads to heavy highly inductive or capacitive loads, and even loads with negative resistance.

It seems that the classic PSU architecture, as in the LM723 standard configuration with emitter follower output, is generally the best behaved, but as you imply, not the best for voltage overhead.

The other area that can cause problems is the point where both voltage and current feedback loops are active. Here a diode AND function seems the best behaved, but I am not sure what the arrangement actually is inside the LM723 (the circuit diagrams shown on datasheets are much simplified versions of the actual chip circuit).

One dodge is to decouple the PSU output to 0V over a wide frequency spectrum using low ESR capacitors, and fit a very low resistance high current inductor between the output transistors and the decoupling.

All good fun, and I have a mountain of dead power transistors to prove it. :(
Navigation
Message Index
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod