pnp vs. npn as pass-device in psu

[JJalling]

Hi,

today i was looking at the schematic for the hp e3631 psu, and one thing puzzled me. (http://circuitslab.case.edu/manuals/Agilent_E3631_Power_Supply_Service_Guide.pdf , page 130).
What is the benefit ( if any) of having the pass-transistor (pnp) on the low side, instead on the high side.

Thanks in advance.

BR Jonas

[T3sl4co1l]

I'd say it has to do with being able to place the current sense resistors either side of ground.

Tim

[uwezi]

It might also have to do with the fact that it is slightly more efficient. Since the pass transistors don't have a high h_FE (TIP36C: as low as 25) having the load current pass through the emitter also utilizes the base current for the load.

Other than that I would also think of the ease of determining the current through the shunt resistors with TL072 opamps with their limited input and output voltage ranges (not rail-to-rail).

[w2aew]

Hi,

today i was looking at the schematic for the hp e3631 psu, and one thing puzzled me. (http://circuitslab.case.edu/manuals/Agilent_E3631_Power_Supply_Service_Guide.pdf , page 130).
What is the benefit ( if any) of having the pass-transistor (pnp) on the low side, instead on the high side.

Thanks in advance.

BR Jonas

It usually comes down to the ability to have less voltage drop from the unregulated side to the regulated side.  With an NPN pass device, the unregulated side has to be at least 1 Vbe above the regulated side (and usually a bit more) so that the regulation circuitry has enough headroom to adjust the base voltage above the emitter which is sitting on the regulated output rail.  With a PNP pass device, the control voltage on the base is 1 Vbe below the emitter, which is sitting on the unregulated side, so there's no headroom issue at all.  Plus, the PNP can be run close to saturation in the worst case, provided the regulator circuit can run enough current through the base to keep the regulation happy.  Thus, with less Vce on the pass device, the regulator can will be more efficient and can tolerate less of a regulator dropout voltage (voltage across the pass device).

The downside is that the PNP pass device is essentially a high impedance output (current source), and the regulator needs to be able to respond cleanly to load changes, which places more difficult requirements on the regulator feedback loop.  With an NPN, the output is taken from an emitter follower, which is a low impedance, making it less sensitive to load changes.

[JJalling]

Thanks to all of you for the explanations.

BR Jonas

[EEVblog]

It usually comes down to the ability to have less voltage drop from the unregulated side to the regulated side.  With an NPN pass device, the unregulated side has to be at least 1 Vbe above the regulated side (and usually a bit more) so that the regulation circuitry has enough headroom to adjust the base voltage above the emitter which is sitting on the regulated output rail.  With a PNP pass device, the control voltage on the base is 1 Vbe below the emitter, which is sitting on the unregulated side, so there's no headroom issue at all.  Plus, the PNP can be run close to saturation in the worst case, provided the regulator circuit can run enough current through the base to keep the regulation happy.  Thus, with less Vce on the pass device, the regulator can will be more efficient and can tolerate less of a regulator dropout voltage (voltage across the pass device).
The downside is that the PNP pass device is essentially a high impedance output (current source), and the regulator needs to be able to respond cleanly to load changes, which places more difficult requirements on the regulator feedback loop.  With an NPN, the output is taken from an emitter follower, which is a low impedance, making it less sensitive to load changes.

You've just explained how low dropout LDO voltage regulators work and why they are unstable, they use PNP pass transistor arrangements.

[sync]

today i was looking at the schematic for the hp e3631 psu, and one thing puzzled me. (http://circuitslab.case.edu/manuals/Agilent_E3631_Power_Supply_Service_Guide.pdf , page 130).
What is the benefit ( if any) of having the pass-transistor (pnp) on the low side, instead on the high side.

I think in the case of the E3631A they done this so the two +/-25V channels have the same ground for the analog control stuff and can share the same bias supply. It's done on other HP power supplies too, e.g. 6236B. If the pass transistor were on the high side then it would need separate bias supplies for each channel and a isolation circuit for the control voltages, like the +6V channel.

[kg4arn]

The downside is that the PNP pass device is essentially a high impedance output (current source), and the regulator needs to be able to respond cleanly to load changes, which places more difficult requirements on the regulator feedback loop.  With an NPN, the output is taken from an emitter follower, which is a low impedance, making it less sensitive to load changes.

You've just explained how low dropout LDO voltage regulators work and why they are unstable, they use PNP pass transistor arrangements.

By chance I was reading this old HP app note:  http://www.hpmemory.org/an/pdf/an_06-65.pdf  and had the very same question as JJalling.

If the PNP pass transistor is on the low side (return current side) as it is in this app note, then the emitter is on the regulated side.  So would the impedance looking into the pass transistor be low?  I thought in the LDOs the PNP was on the high side and the collector faced the regulated output which is a bit different than placing it on the low side.  Am I misunderstanding?

[w2aew]

The downside is that the PNP pass device is essentially a high impedance output (current source), and the regulator needs to be able to respond cleanly to load changes, which places more difficult requirements on the regulator feedback loop.  With an NPN, the output is taken from an emitter follower, which is a low impedance, making it less sensitive to load changes.

You've just explained how low dropout LDO voltage regulators work and why they are unstable, they use PNP pass transistor arrangements.

By chance I was reading this old HP app note:  http://www.hpmemory.org/an/pdf/an_06-65.pdf  and had the very same question as JJalling.

If the PNP pass transistor is on the low side (return current side) as it is in this app note, then the emitter is on the regulated side.  So would the impedance looking into the pass transistor be low?  I thought in the LDOs the PNP was on the high side and the collector faced the regulated output which is a bit different than placing it on the low side.  Am I misunderstanding?

Yes, when the emitter is on the regulated side, then its not really an LDO, since the base has to be driven beyond the regulated rail. Also, it doesn't suffer from the stability issues that are possible when the collector is on the regulated side. This statement applies in both cases (positive and negative supplies).