Request for feedback on circuit design

[gnif]

Hi All,

I am fairly new to all this so please don't be too harsh . I am building a circuit that is capable of displaying 8 digits on 2 multiplexed 7 segment character displays, all using TTL, I do not want to use a MCU as this is a learning exercise. Most of the design is based on parts I have on hand, but suggestions are welcome on better parts to use. This schematic is obviously not complete, I am looking for feedback on if the general design will work as I expect it.

For reference, the display modules are NFD-3641AG-11, I have not yet figured out how to put the part name into the library part, still learning to use EagleCAD. I know it would likely be easier to just use mutliple single digit 7 segment displays, but where would the fun be in that .

EDIT: Just realised I left out the resistors on the transistor bases, and the BC547 might be too small for the potential current for each segment, considering BC338 instead.

EDIT2: Updated drawing with suggestions and added power nets

[Paul Moir]

Few little things I see:

TTL driving 4000 series CMOS  - you'll want pullups on your TTL outputs to drive them all the way to 5v.  CMOS doesn't like low input voltages which are typical of TTL outputs (check datasheets).  TTL is usually somewhere around 3v output voltages.  They really rely on the current coming back from the inputs on the next TTL gate.
4000 driving TTL - 4000 series have super weak outputs (check the datasheet).  You might not be able to fan out as much as you have, which is to say drive both a TTL input *and* provide some current for a transistor.
NOTE:  I think a 74HCT4017 would solve most of these problems for you.  Can't you replace that other CMOS chip in there too?

No base current limiting resistors on your transistors.  I learned that one when I was a kid and all the transistors on my robotic arm popped.  Don't make my mistake!  I had the weak excuse that datasheets were hard to get then.  Mind you, with the el-wusso CMOS outputs you won't be frying anything, but if you get a better driver you'll need to get the output up enough to turn on those inverters too. EDIT: your edit.

Unused inputs must be tied high or low.  Floating inputs actually do cause problems.

Frankly I haven't looked at the logic, so you're on your own there.    Hopefully this is a good start for your exercise.
 

[gnif]

Few little things I see:

Great

TTL driving 4000 series CMOS  - you'll want pullups on your TTL outputs to drive them all the way to 5v.  CMOS doesn't like low input voltages which are typical of TTL outputs (check datasheets).  TTL is usually somewhere around 3v output voltages.  They really rely on the current coming back from the inputs on the next TTL gate. 4000 driving TTL - 4000 series have super weak outputs (check the datasheet).  You might not be able to fan out as much as you have, which is to say drive both a TTL input *and* provide some current for a transistor.
NOTE:  I think a 74HCT4017 would solve most of these problems for you.  Can't you replace that other CMOS chip in there too?

I am just using parts out of my junk bin, but prepared to order some bits, I did not like the idea of mixing TTL and CMOS either, but figured it would have been fine... obviously not :S.

Unused inputs must be tied high or low.  Floating inputs actually do cause problems.

Yeah, that was what I said to ignore ... I have not yet finished the drawing, just looking for glaring problems like the TTL/CMOS issues and missing resistors on transistor bases.

Frankly I haven't looked at the logic, so you're on your own there.    Hopefully this is a good start for your exercise.

Thanks, I am pretty certain the logic is sound as I have prototyped parts of this circuit already on breadboard, just was a little worried about clock timing problems more than anything.

[rs20]

A couple more things:

Those BJTs have no resistors on the base. Since the Base-Emitter junction behaves, loosely speaking, like a diode, you're putting 5V (or whatever) across the diode, and potentially a *lot* of current is going to flow. Unless the 4017 has some current limiting functionality I wasn't aware of. I recommend using N-FETs (the circuit is correct as is for N-FETs), or add appropriately calculated resistances to the base of the BJT.

You're only turning one out of the 8 digits on at a time. This is fine, but I'd also consider turning one two at a time (one from each module). This will double the brightness of the display, and would save 4 transistors and 4 not gates -- but you would need to use a second 4543. Arguments can be made either way, up to you.

Other than that, it looks good to me.

[gnif]

A couple more things:

Those BJTs have no resistors on the base. Since the Base-Emitter junction behaves, loosely speaking, like a diode, you're putting 5V (or whatever) across the diode, and potentially a *lot* of current is going to flow. Unless the 4017 has some current limiting functionality I wasn't aware of. I recommend using N-FETs (the circuit is correct as is for N-FETs), or add appropriately calculated resistances to the base of the BJT.

This was already pointed out my my edit .. I had just updated the drawing with BC338s and appropriate current limit resistors. The display can pull up to 100mA per segment, so 700-800mA max, the BC547s are a little whimpy for that. I might switch out for N-FETs as I have never used them before, good excuse to play with some.

You're only turning one out of the 8 digits on at a time. This is fine, but I'd also consider turning one two at a time (one from each module). This will double the brightness of the display, and would save 4 transistors and 4 not gates -- but you would need to use a second 4543. Arguments can be made either way, up to you.

I had already thought of this and was going to prototype it and see how much difference it really made before I bothered with it, brightness is not a huge worry here as it will be used in a dim environment.

Edit: Just realised also that the transistors would have to be even beefier if driving two digits each as the potential current would be doubled. I do realise that this is pulsed so it is not very likely to be a problem, but I like the belt and braces approach .

Other than that, it looks good to me.

Great, thanks for the feedback.

[Paul Moir]

Regarding the 4017, you wouldn't happen to have a 74LS138 and a 4 bit counter like a 74LS93 would you?  The outputs would be inverted but that should just mean you have to relocate your inverters, which is a good thing because they can supply more current than the 74LS138.  (Traditional TTL outputs can only source a little current, but can sink a fair amount.  Modern CMOS parts like HC can source and sink a lot.)

EDIT:   N-FETs would get you around the wimpy 4017.  Likely you would just need pullups on it's inputs.  They are definitely worth learning about, especially now that low voltage gate ones are available.  They are very handy devices when you want to control a lot of power without spending a lot of current in the (gate/base) drive.  Check out what happens to the hfe on those BC337s when you try to run 800ma of Ic.  That ain't gonna work.

However I am trying to figure out your current budget here.  The 50 ohm resistors suggest you want to limit the current to each segment to something like 50-100ma, but the 4543 can only supply 1.4ma when running at 5v?  So you're only ever going to see 12ma through the digit transistors.

[gnif]

Regarding the 4017, you wouldn't happen to have a 74LS138 and a 4 bit counter like a 74LS93 would you?  The outputs would be inverted but that should just mean you have to relocate your inverters, which is a good thing because they can supply more current than the 74LS138.  (Traditional TTL outputs can only source a little current, but can sink a fair amount.  Modern CMOS parts like HC can source and sink a lot.)

I had to come down to Sydney for other things so made a trip to Jaycar and bought some parts for this, I decided to redesign to drive two digits at once. I also like the idea of using a demultiplexer and 4bit counter, I did not much like the decade counter.

I picked up:
  74LS241 to replace the hex tri-state drivers to reduce the part count by two chips.
  74LS139 instead of the 138 as I will be driving two digits at once now (I did grab some 138s for stock too anyway while I was at it)
  74LS161 4bit counter

EDIT:   N-FETs would get you around the wimpy 4017.  Likely you would just need pullups on it's inputs.  They are definitely worth learning about, especially now that low voltage gate ones are available.  They are very handy devices when you want to control a lot of power without spending a lot of current in the (gate/base) drive.  Check out what happens to the hfe on those BC337s when you try to run 800ma of Ic.  That ain't gonna work.

What part wold you suggest as a basic common low voltage device to play with? What I know of electronics is limied to BJT as that was what my dad passed on to me from his understanding, which never progressed into MOSFET technology.

However I am trying to figure out your current budget here.  The 50 ohm resistors suggest you want to limit the current to each segment to something like 50-100ma, but the 4543 can only supply 1.4ma when running at 5v?  So you're only ever going to see 12ma through the digit transistors.

There is no budget here, it is a learning project and I am happy to throw some cash into it. At some point I plan to obtain a basic CNC router I can isolation route some PCBs for the final design.

When I get home I will redesign my circuit with the new parts.

[Zero999]

Those BJTs have no resistors on the base. Since the Base-Emitter junction behaves, loosely speaking, like a diode, you're putting 5V (or whatever) across the diode, and potentially a *lot* of current is going to flow. Unless the 4017 has some current limiting functionality I wasn't aware of.

The old CD4017 has such weak outputs it can normally drive BJTs or LEDs with no current limiting, when run off 5V. The problem is, when the outputs are dragged down to 0.7V maximum, they can't be used to drive any other logic loads.

Don't try this with the newer HC or AC series ICs.