SparkFun Forums

[thismaechler]

hello,
I'm currently developing a sequencer based on midibox.org / ucapps.de hard/software solutions and button pads / RGB-LED's from sparkfun, I also wrote a Button/LED matrix driver for MIOS (MIDI-Box operationg system).

Now I'am occupied with the hardware side of the BLM. MIDI-Box uses 74HC595 shift registers to drive LED's, now there seems to be confusion about how much current a single output of a 74HC595 can source:

All the datasheets from different manufacturers specify 35mA max. source current and 20mA max. sink current per output. The number that is the reason for the confuction is "max. VCC or GND current 70mA": is this per logic unit / ouput or for the whole device/IC?

I contacted the technical support of ON semiconductors, they say it's for whole device. But fairchild semconductors, which I also contacted, tell me it's per unit (so the max. supply current would be 8x70mA).

I'am a bit confused, is it possible that different manufacturers of the same device type do not fit the same specifications? The numbers in the datasheets are excatly the same, only fairchild semiconductions adds "per pin" to the max. current through Vcc/GND:

DC VCC or GND Current,
per pin (ICC) ±70 mA

Thanks for your help
Matthias Mächler

[leon_heller]

70 mA is the max. current into the Vcc pin and out of the Gnd pin. You will therefore only be able to use a maximum of two 35 mA outputs. Those are absolute max. values, so you should use a *lot* less. You also need to observe the 500 mW total power dissipation limit.

Leon

[thismaechler]

But fairchild semconductor's technical support, which I also contacted, tell me it's per unit (so the max. supply current would be 8x70mA).:

https://fairchildsemi.custhelp.com/cgi- ... rch_thread

leon_heller, do you have personal experiences concerning my question? or did you just read the number in datasheet? I did some tests with the 74HC595 concerning the max. in/out current, if you are interested, you can read more about it here:

http://www.midibox.org/forum/index.php/ ... #msg115545

[leon_heller]

The NXP data sheet I looked at was quite explicit - 70 mA max for Vcc and Gnd. By "per unit" Fairchild would have meant "per chip". An HCMOS chip that is specified to deliver 8x70 mA from the output pins simply doesn't exist! Just because you got a chip to work outside its specification doesn't mean anything and anyone who does that sort of thing for a design that is going into production is being silly, although it might be OK for a hobbyist making a one-off who isn't concerned with reliability.

The TPIC6595 can supply 8x250 mA, if you want to drive high-current LEDs directly.

Leon

[thismaechler]

By "per unit" Fairchild would have meant "per chip".

my question to fairchild tech support:

Is it possible to draw 20mA each output pin (at the same time) from the 74HC595? My aim is to drive LED's.
I read in the datasheet "DC VCC or GND Current, per pin (ICC)", this indicates that 20mA each output will exceed this limit, or is this value per driver / logic unit?
I did some tests regarding input/output current with the 74HC595, which aproved that the 74HC595 is able to provide this source, but I don't know if it will get damaged under those conditions. My tests:

8 outputs high, 20mA output current each (220R -> Ground), Vcc = 5V -> Temp. after ca. 5min: 36° C

I could even get it to source much hihger currents (3 outputs grounded -> 55mA each, temp: 45° C).

Will a supply current value > 70mA just affect the logic levels, or will this harm the device physically? Will a design like this (8 x 12mA out)

http://www.arduino.cc/en/Tutorial/ShiftOut

damage the device?

Thanks a lot
Matthias Mächler

fairchild's answer:

Replied By: MH Eng
Response: Hi Matthias,

The max rate of DC output current and the DC Vcc and GND current is only for one pin.
Thus, if the DC Vcc current per pin is 70mA max, so 8 X 70mA = 0.56A ( max ). In your application, there are only 12mA per pin, 96mA would not damage the part.

[leon_heller]

He seems to think that you can get 0.56A out of the chip! I make that 2.8W, the absolute maximum is 500 mW.

With 70mA current total, the dissipation is 350mW.

Wuth 96mA current total, the dissipation is 480 mW.

FAEs aren't infallible.

Leon

[thismaechler]

From ON Semiconductor 74HC595 Datasheet (maximum ratings):

PD Power Dissipation in Still Air, Plastic DIP† 750 mW

From Fairchild Semi 74HC595 Datasheet (max. ratings):

Power Dissipation (PD)
(Note 3) 600 mW
S.O. Package only 500 mW

In my eyes your calculations are wrong. You assume a voltage of 5V, you multiply it by 560mA, you get 2.8W.
But not the whole 5V Vcc will drop between Vcc and the outputs if you have a consumer at the output (LED, resistor etc.)
You only have to involve the voltage drop between Vcc and output pin, and the current. The voltage drop is only 5V if you short the output pin to ground. At this point, the output driver is in full saturation, it's resistance is the highest then.

I tested this, and you will read a current of 51mA each output if you ground all the outputs (set to high, Vcc=5V). This sums to 408mA -> ca. 2W, you are right you can't get 500mA out of it, and you can't get out 51mA each driver without exceeding PD limit.

But if you connect a consumer to the output, the bigger part of the 5V will drop at the consumer and not between Vcc and the output(s). Then you have to multiply this voltage by the current you read.

This showed up in other tests i did: For 20mA each output, I had consumers with R=220ohms (resistor, each output). 5V/20mA = 250ohms -> 30ohms output driver resistance.
(30R/250R) * 5V -> 0.75 V. This is the voltage drop.

Now multiplied with the total Vcc->output current:
Ptot = 0.75V * 20mA * 8(ouputs) = 120mW.. far below 500mW or even 600mW or 750mW.

I measured the temperature of the device after ca. 20min letting it source 8x20mA, it was ca. 36°C, this makes sense as "†Derating — Plastic DIP: – 10 mW/_C from 65_ to 125_C".

You can't get out 560mA at 5V (maybe after the device got very warm), but you can get out quite more than 20mA x 8, without hitting the Pd limit. 35mA source current is a number that makes sense, if you involve the raising output driver resistance @ higher currents. So I assume you can get out max. 35mA x 8 = 280mA. This would imply a maximum voltage drop of 500mW / 280mA = 1.785V (-> output driver's R=50ohms) without exceeding 500mW power dissipation.

[Philba]

To be fair current and power ratings are separate. Look at the ULN line to see an example where this is more obvious. You can punch a lot of current through lots of devices but the package can't dissipate the heat. I've done a fair amount of experimentation with the 595 (fairchild and one other I don't recall) in both SOIC and DIL packages.

I believe that the 70 mA spec is clear. Yes, it's one pin but when all the current goes through that one pin, it's kind of for the whole chip.

That said, I think the 70 mA number is ultra conservative. I did the following experiment: 8 LEDs, set up for around 11 mA each. Clocked in 8 1's, loaded the output reg and left it run all night (8 hrs or so). That's 88 mA. Temperature rise of the 595's package was on the order of 5 deg above ambient. I haven't experimented beyond that but I did observe some voltage drop. I'd take a look at that before deciding just how much current you are comfortable pushing through our lil friend.

By the way, watch out for surge on the 595's. The inputs blow pretty easy. I was blowing them out until I put current limiters on the inputs (1K series plus a large value - 500Kish to gnd). This happens if you go from all the outputs off to all on.

[thismaechler]

I believe that the 70 mA spec is clear. Yes, it's one pin but when all the current goes through that one pin, it's kind of for the whole chip.

Look at this test result: one output shorted to groud, you read a current of ca. 68mA. This could be the 70mA limit (a little current goes throug the logic stages to GND). 35mA source current max. each output could be the limit to not exceed max. total power dissipation.

If this is true, you could let one single output source 70mA, a power dissipation of 350mW would be the result. For all outputs at once, the 35mA is the magic number, as beyond this, power dissipation limit is reached.

At 5V Vcc, you will *not* get 70mA * 8 out of the device, if all outputs are grounded, I read ca. 56mA @ one output, the voltage drop Vcc->out is 5V (output shorted to ground), this exceeds the max Pd.
But you can get out 20mA each without hitting the max. Pd (Vdrop = ca. 0.75V -> Ptot = ca. 120mW).

So fairchild's answer is a bit fuzzy, but not a wrong info concerning my question. I just wanted to know if I can let the 74HC595 source 8x20mA.

[romani]

There seems to be confusion about how much current a single output of a 74HC595 can source:

All the datasheets from different manufacturers specify 35mA max. source current and 20mA max. sink current per output. The number that is the reason for the confuction is "max. VCC or GND current 70mA": is this per logic unit / ouput or for the whole device/IC?

I contacted the technical support of ON semiconductors, they say it's for whole device. But fairchild semconductors, which I also contacted, tell me it's per unit (so the max. supply current would be 8x70mA).

I'am a bit confused, is it possible that different manufacturers of the same device type do not fit the same specifications? The numbers in the datasheets are excatly the same,