ULN2803 - 0.6v difference

[Callum]

Hi all,
I am constructing a basic 18X / ULN2803 circuit with an infrared remote to turn the ULN outputs on. The IR receiver and associated code is working great but I am having a major gripe with the ULN driver... as we know this is a transistor driver chip, so I get a 0.6v difference to GND. This is bad because this output is actually manipulating the input to another ADC and 0.6v gives the wrong result. (No, I can't just re-configure the other ADC.)

What method can I use to make sure this line sees 0.0v with the ULN output active? (By the way, the output needs to tristate when it's not active, this is why I am using a ULN instead of directly using the 18X outputs.)

 

[Dippy]

As you have said, this is a transistor driver i.e. a junction between 'out' and ground.

When things have ground references it is not advised to use 'low sided' switching.
But do be aware that the "0.6V" varies with load and could be a lot higher according to Data Sheet.
I don't know under what conditions you got your 0.6V.
A simple signal transistor under low-Ice will usually give a lot less than that.
Anyway, doesn't matter, you want zero.

If you have to use low sided switching but want to retain a fairly good OV then consider MOSFETs.
As long as the top-side resistance/impedance is high then the low res of a MOSFET will give a fairly good zero.
Best way is to redesign circuit to give a pukka ground.

I don't know your circuit so can't suggest anything other than this which is based purely on guess and assumption.

 

[TMeyer]

BV:

Coincidently, I am waiting for parts to arrive to allow construction of a circuit similar to yours (IR, ULN, etc). Researching the terms “tri-state” and “low-sided” will be of worth.

Dippy’s comments suggest that in this application, MOSFETs perform a like function to that of the Darlington- driving a medium power device. Similar to statements in Manual3 pages 6/7. One includes a 4001 when doing so as I understand.

Any additional thoughts or information you can share as this is resolved are appreciated. And yes, I am watching the current threads on MOSFETs, relays, etc.

Many thanks,

Terry
Project Type: - Hobby
PICAXE experience level – 2nd project
General Electronics experience level – Self taught
The type of PICAXE chip you are using -18x
Type of board – Breadboard to S/S PCB
Program cable - USB
Editor version – 5.21

 

[Dippy]

Without knowing your requirements or your exact proposed circuit it is impossible to say. And note the 2803 includes 'freewheeling diodes' in the device.

Again , whether or not you need diodes depends on your App. Are you driving an inductive load for example?

Bigvalve says "....to turn the ULN outputs on", so we're driving along in Crystal Ball country...

My suggestion about MOSFETs is that, basically, they act like a resistance. V.high when off and pretty low when on. Whereas anything being switched by a BJT or Darlington will have to go through a semi junction.


For precision, I'd try and rejiggle the circuit design if poss.
These are just some things I am suggesting to take into consideration.
I'm sure you'll resolve it.
All the best.

 

[Callum]

"....to turn the ULN outputs on", so we're driving along in Crystal Ball country...

Ha, very presumptuous of you... however I do know what I am doing when it comes to basic interfacing and have used ULN chips and transistor/mosfet drivers many times before, and my statement is valid, in that I am turning the junction in the ULN on in order for a path to ground to be active, you may have chosen a different term but mine is still correct.

Thanks for all the ideas so far and MOSFETs seem like a good solution as long as they don't drop more than perhaps 0.15v when on.

I know I didn't provide much of a circuit description but there isn't much to describe... an ADC line is pulled to 5v inside a peice of equipment and I am simply grounding this line either directly or through different value resistors through the outputs of the ULN, creating a voltage divider, and a certain voltage on the ADC's line. I can use different resistors to get the voltage I need, but cannot achieve a proper 0v when the line needs to be grounded, only at around 1 milliamp. (this line is also daisy-chained to other devices, hence why it needs to be high-z when I am not controlling it.)

Doubtlessly this has just confused people more which is why I stuck with the simple explanation.

Thanks also dippy for reminding me about the diodes in the 2803... I had forgotten that point.
I'll give a MOSFET a try and let everyone know how it goes

Thanks for the help

 

[BeanieBots]

You might be better off disconnecting the source rather than trying to pull it down to 0v. Use something like a 4066 analogue switch.

 

[Dippy]

So, you have a bit of 3rd Party equipment, which has an 'output' which is taken to 5V.
You load it with various resistors, to get 'certain voltages'.
When you don't need it you want it at high impedance (i.e. as though your section was disconnected). And the junction of load res --- 'Ground switch' is your ADC ground.
Is that right? Sounds unusual. I must have got that wrong as it doesn't make sense to me.

BB's idea of the 4066 looks good.
Four FET switches in a neat package.
About 50 Ohms per switch.

So, you do the calcs. What value resistors are you using?
Work out the voltage: +5V ---- Load resistor ----X---- 50 Ohms --- Ground.
Calc for X. Easy.

You may be able to get 'better' ics than the 4066 as it was invented in 1834.
But at the very least BB has given you a good start where you can now search for similar components.

 

[Callum]

Is that right? Sounds unusual. I must have got that wrong as it doesn't make sense to me.

That's right. And it's unusual, but makes sense in the original application; it is to connect buttons on the steering wheel of a car, to remote control the audio system. Comes from the factory on Holden Commodores. It only needs 1 wire for six buttons, ground is the vehicle chassis. And the button end only needs 6 switches and 5 resistors. The lowest-order switch doesn't go through a resistor, it just goes straight to the common ground... which is where I run into my problems having a voltage difference only on this pin.

I'm adding IR control to this line; which is why I need high-z when it's inactive, otherwise the original buttons won't work as well.

I have stacks of 4066s... I'd need 2 of them to get the 6 required lines, but it could work.