NVMOS amplifier

[ilimzn]

Workhorse, here I am home again, I see the thread has been busy

First of all, regarding changes of the bootstrap cap with a voltage source, I was just using that as an experiment. Besides, it would not change anything as far as driving goes as using a bootstrap cap is exactly for the reason that you are impersonating a voltage source with it. I am still not sure why you would lose headroom on the low end if it was one resistor, but the question is academic as the driver has other issues. Fortunately darkenfriz has provided a solution to it, and more:
With the original circuit:
1) potentially very serious, if the driver stage fails to provide the driving current for any reason, the outputs are fully biassed ON (= lots of smoke!).
2) driver LTP must be able to swing down very close to -V rail if you want to switch off the lower transistor, which requires a very low voltage reference for it's CCS. At 100 deg C for high threshold MOSFETs you are looking at close to 2V up from -V, this could have been a problem. The current mirror approach solves this.

Darkenfriz, I would say that the top MOSFET in the original schematic was also a current cource (disregarding the PSRR issue for now) as it's driving voltage was locked ONLY to the driver stage current and not to the output voltage, by virtue of the bootstrap cap. In other words, the output impedance of the top half is also very high. It is certainly true in you implementation with current mirrors.

I still have an issue with the bias generator. I think Darkenfriz is on the right path as the IRF610 will definitely do a good job of thermal tracking. However, the circuit has gain from Vgs of the thermal sensing MOSFET to the output MOSFETs Vgs, equal to Rgs _outputs / 2*Rgs_ccs. Thermal tracking depends on this 'gain' as the thermal gradient of Vgs threshold is also amplified. This is always the tricky bit with such topologies.
I once had such a problem and this is how I solved it:

1) You need the voltage on the CCS current sense resistor (the one across IRF610 G and S) to track equally with the bias Vgs of the output MOSFETs. In order to do so your MOSFETs need to have proportional Vgs threshold, or the tracking will be off. The simple solution is to aim for equal threshold, which means you need to chose the sensing MOSFET that is built on the same cell technology as the output MOSFETs. In this case, gain will be ~~1 and Rgs_ccs ~~ Rgs_output/2. In other words: carefully evaluate the MOSFETs to use and adjust the CCS 'gain', you cannot just use any resistance ratio.

2) Consider adjusting the idle current by adjusting the resistor* that feeds the junction of the base of the BJT and the drain of the IRF610 in the CCS. Increasing the current requires increase of the IRF610 Vgs, which provides the means of adjustment. At the same time, the current sensing resistor remains the same, and so does the Vgs 'gain' of the CCS with respect to the output MOSFETs.

*At first glance one would expect a constant current source to be a better choice in this place, but in fact a simple resistor has the ability to compensate changes of MOSFET Gm with Vds, which apply equally for the CCS MOSFET as for the output MOSFETs. In order to use this 'feature', MOSFETs built on the same cell technology have to again be chosen.
You also want to reduce idle current modulation by power supply droop. In order to do this, split the adjusting resistor into two and connect the midpoint to -V rail through a large cap (in other words, add PSRR to the CCS by bootstrapping it's reference)

[ilimzn]

One more thing, I was just looking at making the design two stage by using a full LTP for the second stage, and a PNP folded cascode to drive the outputs but the bias generator on that requires some inspired analysis...

[darkfenriz]

Originally posted by ilimzn
Darkenfriz, I would say that the top MOSFET in the original schematic was also a current cource (disregarding the PSRR issue for now) as it's driving voltage was locked ONLY to the driver stage current and not to the output voltage, by virtue of the bootstrap cap. In other words, the output impedance of the top half is also very high. It is certainly true in you implementation with current mirrors.

Yes, you're right.

[ilimzn]

What, no comment to the small book I wrote above re bias generator???

[Workhorse]

This is the new schematic with folded cascodes implemented to achieve good high frequency response.
Bias generator is absent , but i will think again to implement it in this type of circuit.

[Workhorse]

hi every body,

here is a new update to the schematic with Push-pull drivers and optocoupled biasing with mosfet as thermal referencer.

hope you all like it, our new creation I will name it "THE GLADIATOR"

regards,
Kanwar

[amplifierguru]

Hi Kanwar

You sure you would not be partaking of some illicit herbs? This is worth doing, right? SIMPLE. I don't think so.

[Workhorse]

Originally posted by amplifierguru
Hi Kanwar

You sure you would not be partaking of some illicit herbs? This is worth doing, right? SIMPLE. I don't think so.

here comes the Guru, welcome to this thread , kindly expand your views in non jargonmatic illustration.

[amplifierguru]

Seriously Kanwar,

This folly of yours is starting to look like 'one with the lot' - great for marketing I'm sure. Let's see, it's got N CHANNEL, it's got FOLDED CASCODE, it's got RAIL to RAIL ( does it still have that?), it's got SIMPLICITY ( no it lost that way back), it's got OPTOCOUPLED bias (coming),....

It's also got complex thermals, electro's in the signal path, and DC offset? It's got THD? It's got slewing? It's got COMPLEXITY.

It's DIY scary is what it is. GLADIATOR indeed.

[Workhorse]

Originally posted by amplifierguru
Seriously Kanwar,

This folly of yours is starting to look like 'one with the lot' - great for marketing I'm sure. Let's see, it's got N CHANNEL, it's got FOLDED CASCODE, it's got RAIL to RAIL ( does it still have that?), it's got SIMPLICITY ( no it lost that way back), it's got OPTOCOUPLED bias (coming),....

It's also got complex thermals, electro's in the signal path, and DC offset? It's got THD? It's got slewing? It's got COMPLEXITY.

It's DIY scary is what it is. GLADIATOR indeed.

IT is still a regular and simple circuit to achieve less thd ,RTR swing, good high frequency response.

RTR swing- is prime importance of every PA amp to achieve max efficiency.
N-channel - for low cost good solution.
Folded cascodes- again for high slew rate
optocoupled bias- for simple yet effective biasing.

Gladiator-its simple name [or you still have the difficulty to even pronounce it well]

All this is a simple , not a complex amp.

Hey man aren't you the Great GURU of amplifiers .
You shouldn't be talking like this on DIY forums.
i still respect you alot but may be there is a little misunderstanding between us due to your misconception regarding certain points in proamps.

regards,
Kanwar