pin compatible 74595

Discussion in 'General Electronics Chat' started by skk, Nov 12, 2011.

  1. skk

    Thread Starter Member

    Mar 15, 2011
    I really messed up, assuming a TPIC6CPW595 was pin compatible with a 74595. It is not. Does anyone know of a pin compatible 74595 that can sinc 100mA? I need a tssop 16 package...
  2. jimkeith

    Well-Known Member

    Oct 26, 2011
    The 74HC595 is easily available, but there is no way it can come close to sinking 100mA.

    In a pinch, two identical devices may be stacked on top of each other to double the output drive, but it would still be hard put to sink 20mA.

    If you are willing to accept a higher Vo (low), the current sinking capability may be perhaps doubled again.

    Increasing Vcc to 6V also provides a marginal improvement.
  3. SgtWookie


    Jul 17, 2007
    Gosh, I hope you don't have very many boards made. :(

    Those TPIC6 ICs have different pinouts because they need more pins to use as heat sinks. The basic function is similar to the 595, but it can only sink current, and the pinout is different.

    I remember about a week ago you created a thread and posted your Eagle .sch file. Unfortunately, you'd changed the part numbers into function labels like "SRCREG" and "SINCREG" instead of adding some text next to the part. Also, you'd used buses all over the schematic, which made it difficult to figure out what signal was going where.

    I'm afraid that I was tired that day, and I gave up on it after looking at it for just a couple of minutes, so I missed all of that.

    Because I was looking at that section, I missed the fact that you have no provisions for bypass capacitors for your ICs.

    You MUST have AT LEAST one 0.1uF/100nF capacitor across the supply pins on EACH and EVERY IC. If you omit them, you will find that the board is either completely non-functional, or you have intermittent, nagging problems that you just don't seem to be able to find the solution to. Additionally, you should "sprinkle" more of these caps around the board every few square inches to keep your power buses quiet.

    Your 7805 regulator requires an 0.33uF cap from IN to GND, and an 0.1uF cap from OUT to GND, and these must be located as close as possible to the regulator itself. You can add larger caps (the 100uF are OK) but you must not omit these small caps; they are required. If these caps are not placed very close to the terminals, then you can wind up having the regulator oscillating at frequencies in the MHz range; and the output voltage will usually be much lower than normal.

    I don't recall that you ever uploaded your .brd file, either. It is a good idea to get some other eyes looking at your project; you are so familiar with it that it would be very easy to overlook something. For example, if you have a really long clock signal path (or other logic signal), you may wind up with a lot of "ringing" on it due to inductance.

    Chances of having a board come out perfect the 1st time, particularly one that's fairly complex for a relative newcomer to electronics, are pretty low. That's why you have prototype boards made, to find out what all is wrong with your design.

    Sorry that we didn't catch all of this before, but there have been a lot of people and a lot of threads going on; you have seven threads going that are all related to this project without links to any of them, so there is a lack of continuity; one really has to do some searching/reading to stitch a mental picture together about what happened when.

    It's OK to have "breakout" threads that are specific to one particular part of the circuit; but you really should link back to the main project thread. It's frequent that changing something in one place will affect something somewhere else.
  4. skk

    Thread Starter Member

    Mar 15, 2011
    Thanks again SgtWookie. I know I have a bunch of similar threads, but this is because I have am on revision 19 of my idea. I have produced 6 different similar families of devices. This one is skk- I went back through my posts on all different devices and found the ones that have gone into my current design, the others were dead ends. I think this is the useful list of parts that went into skk-

    The caps you suggested (.33uf and .1uf) for the voltage regulator are new to me. I see they are what the manufacturer recommends, but I have been using this 100uF 7805 regulator circuit for like 5 years now with no (apparent ) noise problems. Why does this work for me at all. I have used many different micro controllers and different logic components all without issue. I have used smt, and thru hole with no problem. My designs power multiple devices, sometimes, off 1 regulator also with no problems. I run from battery or from transformer. I don't get it. I will defiantly take your advice and add these caps, but why does it work for me at all? I suspect probably because all the IO signals come from bouncy switches anyway, and everything else just drives LEDs, or is inside the uController.

    Should I get rid of my 100uF caps? What are they doing?

    2nd issue, decoupling caps. I have none because I have no noise problems, at least so far. However this latest design has doubled the number of latches and added shift registers, so sounds like a good precaution. I think I want to use smt parts, what type would be the best / most cost effective?

    I will upload my board and updated schematic when I get the new shift register drawn and the caps in.

    I have a question about this stacking two parts on top of one another: Has anyone actually make this work for a TSSOP-16? I think I am pretty good at hand soldering smt, but I wouldn't even try that...
    Last edited: Nov 16, 2011
  5. SgtWookie


    Jul 17, 2007
    I see.
    This one is missing a digit; "6090" points to an old, completely unrelated thread.

    Perhaps pure luck?
    The oscillations only occur under certain conditions if there is not enough low-ESR capacitance near the regulator. These caps are specifically recommended by the manufacturer to prevent the oscillations from occurring. If one doesn't follow the manufacturer's recommendations, don't expect the circuit to function as specified by the manufacturer, particularly when there are specific warnings about certain behavior as this.

    I don't get why you have not experienced problems either - unless your circuits are VERY low frequency.

    I really don't have enough information about your circuits to tell you WHY they are working anyway; I have yet to see a board that you have created. However, I CAN tell you that we very frequently have new members that are having difficult-to-explain problems with their circuits, and it turns out that they did not include bypass capacitors; most of the time adding the caps takes care of the problem.

    [/QUOTE]Should I get rid of my 100uF caps?[/QUOTE]
    No, keep them.
    They greatly improve the transient response. You're going to be switching several hundred mA current on and off quickly; that'll create rather large transients.

    You'll have to price them out. MLCC's (multilayer ceramic caps) would be fine. You can get 0.1uF's in as small as an 0402 package, but that's too small for me to deal with. 0603's the smallest I deal with anymore. 0805's are much easier. The larger they are, the more expensive they get.

    You can get X7R 0603 caps at Digikey for under $0.02/ea when you're buying 100.

    OK, good deal.

    I've never seen SMT/SMD IC's stacked. You usually do that with just DIPs. I've stacked SMT/SMD caps before where physical strength wasn't an issue, as the part was going to be filled with Eccofoam anyway.
  6. skk

    Thread Starter Member

    Mar 15, 2011
    I went back and breadboarded this design. My current source is a single pin of the mm74hc595 ( per 7 segment LED ) driving a 2N2907 giving me 350mA max. The current sinc is a tpic6c595 giving me 100mA drain per pin or 500mA for max for the device. My only voltage source ( VCC ) is +5V. The 7-segment display being multiplexed 1/3rd of the time is a HDSP-C1A1 common anode display:

    Since the usual forward voltage for the 2 leds is 4V, I have a bunch of 51 Ohm resistors for current limiting resistors on each segment of the 7-segment display. Assuming 20mA per segment, and rounding the resistance up.

    Except whoops, the LEDs are only on 1/3 of the time so I need more current. I assume I need 60mA per segment on 1/3rd of the time to achieve full brightness. However, this is how I hooked it up, with 51 Ohm resistors, and the brightness is fine. How can this be?

    I put another 7 segment next to the others and just hooked it up directly to vcc and gnd with no current limiting resistors at all, and it did not appear much brighter than the multiplexed ones, certainly not 2/3rds brighter.

    What am I missing here, is this just a trick of the eye?

    If I increase the current limiting resistors do I save power or am I just converting it to heat in the resistors?

    And as for the base resistor on the transistor. If I have a 51 Ohm resistor on each of the led segments, that would give me a max current of 20mA * 7 + 10mA or 150mA. Do I even need a base resistor?

    Also with the LEDs, do I use the max forward voltage or the usual forward voltage to compute the current limiting resistors?

    Thanks again in advance...
  7. SgtWookie


    Jul 17, 2007
    If you go by the standard Ib=Ic/10 rule, then you're sinking 35mA using the MM74HC595, which is their absolute maximum source/sink current. I don't recommend operating them at their absolute maximum. ±20mA would be more like it; which would mean 200mA Ic for the 2N2907's; unless you want to start qualifying them individually to make certain the Vce(sat) won't be excessive with the lower Ib.

    OK. Your maximum current limiter is still the 2N2907. It's going to have more Vce due to Ib being lower than it might.

    You might consider using TWO outputs at a time from the 74HC595 to drive a base simultaneously. You could safely get 40mA base current that way. Use separate resistors from the two outputs to the base; you really don't want to directly connect a pair of outputs together if you can avoid it.

    OK, you're good so far.

    If you look at the datasheet again, the absolute maximum current for 100% duty cycle is 25mA/segment, and for 10% duty cycle, 80mA/segment and 0.1mS duration. If you tried running it with 60mA current for 1/3 of the time, you'd probably shorten the life of the displays.

    Did you measure the actual current with the resistors, and without the resistors?

    It's less current total, so it's less power expended. Some of the power will be dissipated in the resistors as heat. Since diode Vf is ~4, and you're dropping ~1v across the resistor, that'll be ~4/5 of the power in the LEDs and ~1/5 in the resistor and Vce(sat).

    Absolutely. In that case, you will need ~15mA base current.

    Use the typical Vf.
  8. skk

    Thread Starter Member

    Mar 15, 2011
    One more thing I don't understand. If I have the current through the LED limited to 150mA by the current limiting resistors on the segments, why do I need the base resistor? Also the value of the base resistor should be > 6.6 ohms correct?
  9. SgtWookie


    Jul 17, 2007
    Because the base-emitter junction "looks" like a forward-biased diode. The Vf of a forward-biased diode starts out fairly low at low current (around 500mV @ maybe 250uA), but that increases as the current increases.

    Here's OnSemi's datasheet for the PN2907A:
    Have a look at page 4, figure 4, "Collector Saturation Region". They show curves for Ic= -100mA and Ic=-500mA. You'll have to extrapolate to get the typical curve for Ic=-150mA; Vce should be somewhere around 120mV when Ic=-150mA and Ib=-15mA.

    If you look at page 2, under "ON CHARACTERISTICS", see Base-Emitter Saturation Voltage.
    It actually shows for Ic=-150mA and Ib=-15mA, the max Vbe is -1.3v.

    Now, for the 74HC595, it doesn't show what the voltage drop will be for Vdd=5v and Io=15mA. Let's just estimate it'll lose 1v.

    So, Rlimit = (Vin - Vbe) / (Ic / 10)
    If you're using 5v for Vdd, we'll subtract that 1v you'll lose in the I/O pin.
    Rlimit = (4v - 1.3v) / (150mA / 10)
    Rlimit = 2.7 / 15mA
    Rlimit = 180 Ohms. That's a standard E24 value:
    See the green columns.

    You could certainly use 200 or 220 Ohms without hurting Vce much.

    Now, let's get back to your LED current limiting resistors. You must be using the red ones, because they're the only LEDs that have typ 4.0V/2.0V @ 20mA.
    The 2N2907 will drop ~120mV across the collector-emitter junction with 150mA current; it'll be less otherwise. The TPIC6CPW595's will drop a little across the drain-source, but not much. You can calculate it by using the Rds(on) figure and multiplying it by the desired current.

    Then subtract the Vce(sat) and Vds from Vcc, then subtract your 4v or 2v typical from that remainder. Whatever's left is what you need to have 20mA current flowing through.

    Yes, it should be ~180 Ohms. 200 or 220 would be fine, too.