AoE 3 (power MOSFETs in parallel)

I'm slowly working my way through AoE 3, lots of excellent stuff so far, and came across 3.6.3, running power MOSFETs in parallel in linear mode. I know this has been discussed here in the past and the consensus was that it can't be safely done with regular vertical MOS devices without ballast resistors or some form of active current sharing.

Fig 3.117B looks like a neat trick, which I may have a use for soon. I pulled up the schematic for the HP bench supply mentioned, to see what they used for the diff amp current sink - turns out it's just a 100K resistor to -12V, that's clearly adequate for the job.

What puzzled me is that in the HP schematic, the NPN devices Q4 and Q5 are connected upside down (c & e reversed), the AoE schematic shows them the right way up. Any ideas why HP did this? I'm assuming it's deliberate.

I have seen BJTs connected this way before to get lower saturation voltage, but there doesn't seem to be any advantage to doing it here. Or am I missing something?

Reply to
RBlack
Loading thread data ...

It is a cute trick.

Yes, I hadn't noticed that on my first pass through the manual's schematic, but I did notice it recently. I've been intending to open one up to see if this was perhaps just a drafting error.

It's called inverted mode. Most transistors have severely reduced beta in this mode, but this circuit doesn't really need a high beta. Inverted mode npn BJTs in this circuit would have a 6 to 8-volt breakdown voltage, so perhaps the HP engineers were thinking of protecting the MOSFET gates? But there don't appear to be any dangerous voltages around.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

What was the other trick to fix transistors that had been zinged with excess reverse Vbe and zenered too many times? Smoke a Marlboro and crush the cigarette butt on the metal TO can? But I guess that's no longer politically correct these days.

A drafting error in an HP schematic? Mon Dieu!

HP boat anchor stuff was sometimes protected with belt and suspender and then another belt.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

The maximum voltage the FET gates could be exposed to is +12V.

Looking at an E3612A instrument, I first saw the two diff-amp transistors are TO-18 can 2N2222A. Whoa! This was a recent vintage Agilent production run, and all the other transistors are plastic, so I thought they must have had a special reason to use these old metal parts, such as a known Veb breakdown voltage. But looking carefully at the traces, whoa #2!

The schematic is in error; it's the collectors that feed the MOSFETs, as we show in Figure 3.117B. Perhaps it's fortunate I that misread the drawing when transcribing it. They have 75-ohm gate resistors (for damping), that I failed to show. The FETs are HV types, which can otherwise oscillate at RF frequencies, when in linear-operation modes.

Maybe, but not in this case.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

That is indeed a surprise. I can't remember ever seeing a drafting error in an HP document. Other companies, yes, lots.

Real men don't use gate resistors, only ferrite and only if absolutely necessary :-)

I always wondered why people use 10A gate drivers with sub-ohm internal Rdson paths and then hang on a 15-20ohm gate resistor. On my last HV driver a month ago I used 10A drivers to muscle around about 5000pF. Sans gate resistors it was a lot more efficient. In linear mode it's often best to thoroughly review the layout for issues and then work with ferrites.

[...]
--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

Wow, I was going to say that I always like a few ohms of gate resistance. That must make me a mouse. :^)

George H.

Reply to
George Herold

Then you have to add ohms for the EMI guys. It's a vicious cycle ;)

Cheers

Reply to
Martin Riddle

I've got a moderately vintage bit of kit, the 7035B X-Y Plotter. Inherited from an acquisition of theirs I think. The design is astonishingly primitive (wirewound resistor position slides!), yet clever, quite different from the "overdo everything" mentality of HP's test gear department. The two inputs are isolated from each other; the power transformer has tons of windings (drive power x 2, circuit power x 2, and HV polarization for the electrostatic "make the paper stay put"); and there is exactly one power transistor (a 2N3055) per channel which operates each PMDC motor. (Hint: there's also a pair of neon lights coupled to CdS photocells on each channel!)

The manual says various parts of it were up to Rev J or so. Geesh, get it together!

Tempted to tear that out and replace it with a proper (discrete, of course!) bipolar servo amplifier that HP would've been properly proud of. More bandwidth too... though I need to make the Y axis drive string a bit more taut or sticky before doing that.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

I had a giant HP X-Y plotter, back in the day, purchased to make printouts for our schematic diagrams and PCB layouts. It made large plots, to 3 by 5 feet or something like that. Impressive, but now it's hard to imagine a need for such a thing. Now we use 1200 dpi laser printers instead. Or an inexpensive Brother 11" x 17" ink-jet to print the larger schematics.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

Lovely! I only use this thing for stuff like plotting transfer functions... one example of which was a UHF oscillator hack, for which the Vout vs. FCV had... nipples, and loops, and hills and valleys... :o)

I still need to pick up a proper office printer, myself. (And no, this thing wouldn't be of much use, not unless I had an HPGL-to-DAC dongle handy.)

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

Brother periodically puts their 11 x 17 inch ink-jet multifunction printers on sale, I only paid about $130 for it, IIRC. They make up for the low price with high-cost ink. We paid $600 for our big Brother color laser printer, but it's cost over $2000 in toner so far. About 10 cents a page, which is pretty steep.

--
 Thanks, 
    - Win
Reply to
Winfield Hill
+1 on the B-size Brother all-in-ones. I have two of them.

Even the newer Brothers work fine with generic ink, *if you peel the ID chips off the old cartridge and put it on the cheap one*. It somehow notices that it's been fiddled with, but still works.

Cheers

Phil Hobbs

Reply to
Phil Hobbs

Hmm.

We have the MFC-J6520DW, which sells for $180 to $230, except when Brother periodically puts it on deep sale. One nice feature, it can also scan an 11 x 17 sheet.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

OK, thanks for confirming that.

I have played around with this circuit a bit in SPICE, but the only lever I have to pull to simulate a mismatch between the two FETs is to force the device temperatures apart. Manufacturing tolerances of V_GS threshold will play a much bigger part I suspect, so I'll just have to breadboard this and try it.

The application is a high-power adjustable current sink (say up to 20A and up to 30V), a one-off for use in our lab for testing batteries, PSU designs etc. Our new test tech blew up my previous design, a very crude effort made with 2N3055s and ballast resistors. Once I get something working I'll post it.

Reply to
RBlack

You could edit the VTO value of one of them. But it's easy to analyze the circuit: Calculate the diff amp's voltage gain, RL/re, then determine dVin for say 1-volt Vgs difference, and hence the current imbalance on Rsense.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

I have the MFC6490CW in the lab, which doesn't sense cartridge identity and works fine with generic ink. I do most of my schematics by hand on B-size vellum, so having a B-size scanner is very useful.

At home I have a newer one, which I think is the same as yours--a B-size duplex all-in-one with 2-sided scanning as well. That's the one where swapping out the ID chip from the top of the cartridge is necessary. It's just glued down, and peels off easily.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

A few ohms are ok, sort of. But more than 10 ohms is like listening to a Mozart concerto in a Dodge Challenger, they'll make you leave the car :-)

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

You need some sort of gate driver anyhow, so one opamp section per mosfet is reasonable. Close a loop on a small source resistor to make each mosfet into an almost-ideal current sink. That works better and takes fewer parts than most other circuits.

--
John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 
 Click to see the full signature
Reply to
John Larkin

Once upon a time (like pre-1980) I had an E-size plotter for plotting out chip layouts for visual checking. I can't remember the brand now, it wasn't HP, but it talked hpgl.

It'd take all afternoon to do a complete plot.

Microrel had a hang-on-the-wall inkjet where the paper was stationary and the 8' wide head progressed from ceiling to floor and was very fast. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
 Click to see the full signature
Reply to
Jim Thompson

We just got a new Sharp B-size digital color copier. It's also networked as a scanner/printer. Unlike some HP beast, it uses toner by the bucket-load. One of our older HP color printers is now running close to $1000 for a full set of toner cartriges. Bye-bye, HP!

I draw schematics on D-size vellum but everything gets CAD'd as B size.

Remember when HP had integrity? It's been a while.

--
John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 
 Click to see the full signature
Reply to
John Larkin

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.