Basic problem with transistors

[giftiger_wunsch]

My circuit uses parts the OP probably already has on hand.

As soon as I get a finalised schematic (I'm sure it'll happen eventually) I'm going to have to order in a load of parts anyway. I have a couple of transistors sitting around but nowhere near enough. I have a grand total of three signal diodes + whatever I can salvage from a discarded VCR's motherboard, so I'll need more of those, 0 schottky diodes... etc. hopefully bulk prices on the transistors, diodes, etc. will reduce the unit prices a fair bit.

 

[giftiger_wunsch]

I got my friend to re-test the current through the motors while in operation. Multimeter showed:

~243mA while moving
~740mA while stalled edit: he tried a few more times and got a peak of 1.05A


I don't know how these values will affect the other components, but it looks like I'll need to get some higher-power resistors, given that my set of resistors are 1/4W.

 

[audioguru]

The motor's current does not flow through resistors so their size is probably fine.

 

[Hero999]

The maximum power dissipation of R5 and R6 is just 58mW so I think your ¼W resistors will be fine.

Just a thought, have you looked into using high gain transistors?

The ZTX690 (NPN) has a maximum saturation voltage of 0.5V with a collector current of 1A and base current of just 5mA.

The ZTX790 (PNP) has a maximum saturation voltage of just 0.45V with a collector current of 1A and a base current of just 10mA.

If you use high gain transistors (at least for the low side) you can eliminate Tr1 and Tr2 in my original circuit.

**broken link removed**
**broken link removed**

The above transistors are also very tough and will probably withstand the stall current. Just make sure the motor can handle being stalled so doesn't catch fire.

A simple 0.5A polyfuse is all that's needed to protect the motor assuming that it stalling could cause it to burn out; it might be all right.

The circuit posted below uses the high gain transistors for Tr3 and Tr4 and normal transistors for Tr5 and Tr6.

If you use high gain transistors for Tr5 and Tr6, R5 and R6 can be increased to 1k which will save power.

At 200mA the saturation voltage of Tr4, plus the turn on voltage of the Schottky should be < Tr3's turn on voltage, so the protection for both inputs being high should still work.

 

[giftiger_wunsch]

It appears that the schematic is slowly creeping back in the direction of my original schematic, with the addition of the high-gain transistors. I'll check out the link you gave me in a sec.

The motor's current does not flow through resistors so their size is probably fine.

Ah... yes, of course not. I forgot that


By the way, the motors were fine stalling when tested, and they won't be intentionally stalled or anything, so the motors shouldn't have a problem with being stalled. I'm more concerned about the high current damaging other components.

 

[Hero999]

It appears that the schematic is slowly creeping back in the direction of my original schematic, with the addition of the high-gain transistors. I'll check out the link you gave me in a sec.

Bear in mind that the high gain transistors are about three to four times as expensive as normal transistors.

This probably won't be a problem unless you're going to mass produce this.

 

[giftiger_wunsch]

Bear in mind that the high gain transistors are about three to four times as expensive as normal transistors.

Hmm, I'll have to decide which is more worth it: more work (and there's already a lot of it) or more money. It may be difficult to obtain the high gain transistors anyway, so that might end up deciding for me. I usually get my electronics supplies at maplin, and I don't remember seeing any high-gain transistors on their site.

 

[marcbarker]

Representative model of the digital output, behaving exactly as officially stated in Rabbit uP Datasheet PDF file DC parametric data, posted earlier. You'll see at the stated "8 mA sink", the output = 0.4V. At "8mA source" the output = 2.4 V, exactly as officially published data. This isn't just using the nearest standard component from the library, (hoping that it's 'near-enough') This is created specifically to represent the DC source resistance at 8 mA drive current, so I don't think anyone can argue that this isn't representantive of the source now!
**broken link removed**


Below, is the 'curiosity' organic 2bjt/2fet circuit, simulated with LTSpice. With 200mA load. Notice that it appears to work well. The BJT is neither in CC or CE configuration and seems to be just-enough driven. If you feel you have to classify it, maybe call it a:'self-bootstrapped emitter follower'. It's not saturated, but does it need to be? With better transistors, the Vce (lost voltage) might be even come out lower than this, i.e. Zetex as suggested by AG. Even with a loss of voltage, in some apps. it's probably OK. Don't forget an electric motor on lower voltage and same current still has the same torque, it just spins slower that's all.

On the left, the green and yellow columns represent the Nicad and Motor voltages respectively. Notice they're + 0.2 V, because of the shottkey diode voltage drop. Red and blue voltages are due to the modelled source/sink impedance of the Rabbit IC. You can also see the voltages of e, b & c of the BJT there too.
**broken link removed**

Now with higher 400mA motor loading. Notice the base current increase proportionally. There is still a useful voltage on the motor. The BJT is dissipating less than 0.2 W, well within TO-92 package rating.
**broken link removed**

With motor stalled. Notice the dissipation of the BJT < 1 W? That is no 'instantaneous destruction' ocurring to any parts! This could mean there would be no need for any protection devices such as polyfuse (which reduces voltage on motor probably more than the transistor!
**broken link removed**

 

[audioguru]

Your SIM program shows that some circuits will work if you spend hours testing transistors to select the "typical" or best ones.

You cannot buy "typical" transistors anywhere. You get good ones and not so good ones. All the transistors will meet the minimum spec's on the datasheet that the SIM program knows nothing about.

So if you want every circuit you make to work properly then design it using minimum spec's. Then this circuit will not work.

 

[giftiger_wunsch]

this circuit will not work.

I consider my parade officially rained on Though I hope there's no such fatal flaw with Hero's schematic? I intend to go with that one, making my modifications to it.

 

[marcbarker]

Your SIM program shows that some circuits will work if you spend hours testing transistors to select the "typical" or best ones.

You cannot buy "typical" transistors anywhere. You get good ones and not so good ones. All the transistors will meet the minimum spec's on the datasheet that the SIM program knows nothing about.

So if you want every circuit you make to work properly then design it using minimum spec's. Then this circuit will not work.

What do you mean by 'properly'? This is a curiosity circuit, not a mass produced product aiming for 100 % yeild!

And it's not just 'my sim program', it's used by many people on this forum, it comes from Linear Technology.

Even if what you say is correct, it doesn't take more than a minute go through a £1-worth bandolier of 4p transistors and pick out the best ones.

Anyway, had you not heard? There's Worst Case Analysis mode available in most "Analogue Simulators", where device tolerances and production spreads are fully modelled, more than enough for any pessimist. A stage further from this is MonteCarlo analysis + WCA, to predict actual yeild. You're welcome to suggest how the simulation can be improved. Or build it and see.

It's diminishing returns chasing 100% yeild for commercial products, that's why aerospace electronics costs 1000x as much, and it still fails in the feild! In the 1990's Iraq conflict, the armed forces used commercial GPS units because the mil-spec ones weren't available on time, probably because they were trying to make them too perfect.

 

[giftiger_wunsch]

The software simulation gives a good idea of how the circuit will work, but either way I'll be going with the suggestion hero put forth, and modifying it with schottky diodes on the ground side to fit the theory I've been discussing with marcbarker. Once I can get a final schematic up and get a couple of opinions on it, I'll order in the components and build an actual model on a breadboard to start with, to test if it will work in practice before exposing my expensive microprocessor to an uncertainty

Simply replacing the parallel port input with a battery I can manually switch should give me an idea of the sort of values I'll actually be able to produce, and how the arm will function in practice. If I can get that working, I can finally wire it up to my microprocessor...

 

[marcbarker]

The software simulation gives a good idea of how the circuit will work, but either way I'll be going with the suggestion hero put forth, and modifying it with schottky diodes on the ground side to fit the theory I've been discussing with marcbarker.

I like Hero999's h-bridge best. And the IC-based one posted before that has less parts. The 'curiosity 2bjt/2fet' one is intriging, in that it works in a novel way.

(Open-forum-mode on) What's the issue with diodes in the ground side?

 

[giftiger_wunsch]

These are annotated schematic diagrams which marcbarker requested to explain my diodes in the ground part of the circuits, perhaps other people can comment on them also.

Note that the lack of resistors / the fact that the BJTs won't work correctly isn't important, I'm just using the basic H-Bridge construct to simplify the diagram and focus the attention on the earth wiring. If this ground-side wiring works the way I expect it to, I can apply the concept to hero's schematic, thus obtaining perfection okay not quite, but it'll be a good composite of the two

Please excuse the rough hand-drawing of the circuits. Red shows current for the uP circuits of M1, blue for the same for M2, and black for the motor driving current. The green in the second circuit shows what would happen if not for the schottky isolation diodes I used, and black in that schematic shows the short-circuit in the battery circuit which would result.

 

[marcbarker]

OK, I see where you're coming from now....

The sketches you'd posted is a concept only to illustrate a point, and not a practical circuit design. A lot of detail is ommitted, to improve clarity.

Looking at the righthand sketch. In this theorectical scheme, the current which activates the BJTs, originates from the IC, flows into the bases and out the emitters, going towards a logic L output where it returns back to the IC. There is no hard-wired ground between uC and battery ground. Presumably this would be to save wiring.

The right hand skech showing both PA0 and PA1 are both = HH. This could happen due to a uC start-up condition. If there was no protection present in the hardware, the H-bridge would go into 'shoot-thru' and that's not good.

The current flows in the green coloured path, however, the right hand diode underneath M2 will block this current from reaching the PA2 and PA3 (labelled M2 gnd). Therefore the transistors are starved of current (because of the diode) and they can't switch on very well, if at all.

So protection is offered, but it costs the voltage drop in the motor circuit, which is probably acceptable.

Like the saying goes "there's more than one way to skin a cat". There are probably more ways to implement shoot-thru protection and this is just one of them.

 

[audioguru]

The software simulation gives a good idea of how the circuit will work

Only if the transistors you use have the same spec's as the simulator's transistors that might be wrong or are "typical".
You cannot buy "typical" transistors anywhere. They are a gamble.
So you should design circuits using minimum spec's then every circuit you make with passing transistors will work perfectly.

I designed many circuits using minimum spec's for parts. One circuit was manufactured thousands of times and only two did not work properly. One had a capacitor installed backwards and the other had its quad opamp IC installed backwards.

 

[giftiger_wunsch]

So you should design circuits using minimum spec's then every circuit you make with passing transistors will work perfectly.

Well I'm not experienced enough to be able to determine how the BJTs will behave, so I'm relying on you guys to help on that front I'm going to try to draw up my full schematic by tomorrow. It will be based on Hero999's schematic, with the addition of the schottky diodes controlling ground flow as shown in my last-posted schematics.

A couple of quick points to confirm with whoever might be able to answer these:

1) The protection diodes for the BJTs were eventually deemed unnecessary, correct?

2) Given that the stall current of the motors was measured at about 750mA, with a peak of 1.05A, will the circuits be able to handle this, or require the current to be limited to avoid destruction when the motors stall? I'm fairly certain the motor will survive okay, just need to double-check that the BJTs, diodes, and especially my microprocessor won't be adversely affected.


Thanks. I'll post the completed schematic as soon as I can.

 

[marcbarker]

|-----------------------------------------------|
|..................................................................|
|..................................................................|

You cannot buy "typical" transistors anywhere. .

|..................................................................|
|..................................................................|
|-----------------------------------------------|

I just had to frame this one!

 

[giftiger_wunsch]

Oh dear, do behave, children. The fact that the quotebox killed your frame is exactly what you deserved there marc

 

[marcbarker]

'Typical' is like marriage, it's better OR worse, but on average it's 'typical'.

Of course one does worst case analysis and tolerancing in a mass-maufactured product. Different companies have different slant on it. Ford cars like to tighen all tolerances. In spacecraft design, the WCA is so pessimistic that the transistors only have a hFE of 10 and parts are operated a fraction of thier commercial rating, and even after all that, every circuit is doubled for redundancy, sometimes twice!