74LS30 Testing Failures

[pjh]

In recovering failed or erratically functioning S-100 boards, one of the things
that I do is chip test as many of the 74xx TTL chips that I can. I use a breadboard
set up with a 555 timer and a 7493 4-bit binary counter to provide input signals for
the chips to be tested. On chips with more that 4 inputs, I pair up the inputs so
that the whole chip can be tested at one time. One thing I have noticed over time
is a high failure rate of the 74LS30 chip, even on orders of new chips from suppliers.
First, let me define 'failure' as I see it in this case. Perhaps someone can
tell me if these are in fact failures. In my testing I check first to make sure that
any chip is functioning as it is designed, and then I watch for clean HIGH-LOW
transitions in the outputs of the chips. The 74LS30, an 8-input positive-nand gate,
seems to 'fail' often in that the output doesn't do a clean HIGH-LOW transition but
instead does a HIGH-DIM-LOW transition. I have been assuming that a HIGH-DIM-LOW
transition in any chip would produce an invalid logic state in a circuit for the period
of time that the DIM state exists. So, I replace the chips that produce these results
as well as any obviously bad chips.
I find these 'failing' 74LS30 chips often on boards that I am chip testing but I
also find that in orders of 74LS30 chips, many of them are also 'failing'. I may
get as many as half or more of the chips I have ordered to produce this HIGH-DIM-LOW
output. Other chips such as the 74LS240 or 74LS244 may show the HIGH-DIM-LOW
transitions as they begin to fail but I rarely see such HIGH-DIM-LOW outputs on
orders of any other new chips. Has anyone else seen such a high failure rate in
74LS30 chips?

Phillip

 

[Hugo Holden]

Phillip,

This is very interesting. I have worked on many pcb's with large numbers of 74 series and 74LS IC's, I have never yet found one sort more prone to failure than another, never noticed the 74LS30 more prone to failure. One thing is, for the most part "relative failures" are rarer. Having said that, one failure I had on my SOL-20 computer..the keyboard would intermittently stop working. It turned out that a 74LS93 had an intermittent output from its first flip flop and the output was disappearing, only from time to time, killing the keyscan circuit.

Most of the time, it is the contacts to the oxidised IC pins in 40 year old IC sockets that give issues (when the IC's are in sockets) rather than the IC itself.

In the 74LS (and 74 series IC's of course) the output stages are good at sinking current to ground, but poor at sourcing it. A good test to see if the output stage is ok is to tie the output high with a 470R to 1k resistor in your test jig. If the output stage is normal, it should be able to pull that below 1V.

The output stage design of a 74LS30 is the same as a 74LS00,04,10,20 so in theory it should behave in the same way.

Hugo.

 

[daver2]

You may also be forgetting that each device has a 'time' element in that the output of the device may only be guaranteed to reflect the input state after a short delay. However, it depends how you are testing it of course. This delay is in the order of 15-30 ns so you will probably never see it if testing manually, but you could do if using an automated way to test.

The other possibility is decoupling on the power supply pins of the IC under test.

Just a few things to think about. If you can post more details of your test system I might be able to help further. For example, how are you reading 'dim'? A multimeter, oscilloscope a lamp?

Dave

 

[Hugo Holden]

You may also be forgetting that each device has a 'time' element in that the output of the device may only be guaranteed to reflect the input state after a short delay.
Dave

I was wondering about this too, but Phillip said the signals he was using to drive the IC under test were coming from another TTL IC (7493), which means they will have standard low to high and high to low transition times, so looking at the output of the IC under test, this should also have standard transition times and in theory should not have an altered or intermediate wave shape between the high and low state on the device's outputs. There will be the usual propagation delay via the gate etc but that shouldn't alter the output transition shape.

Good suggestion about checking the supply decoupling.

 

[Hugo Holden]

The 74LS30, an 8-input positive-nand gate,
seems to 'fail' often in that the output doesn't do a clean HIGH-LOW transition but
instead does a HIGH-DIM-LOW transition.

Phillip

Just thinking about this for a while I wonder if the new IC's you have been sold are not in fact genuine 74LS30's That sort of intermediate state at the output suggests there are two output devices, like a pair of mosfets which are acting in an analogous manner to an analog output stage with cross over distortion. In other words there is a time period where neither device is conducting in between the high low transition. I wonder if you have been sold some chips that are in fact cmos, like say a 74HCT30 but are labelled as LS types. Or they could be just bad fakes/clones.

I'm very suspicious of what are said to be new 74 series IC's as they have been obsolete for so long. I only buy vintage date code 74 IC's that look genuine, TI,National,Signetics & Hitachi & Motorola are very good. The latter two are hard to fake because genuine ones have very rounded edges on the packages unlike most other brand. National IC's tend to have a unique round indentation too. One way to be nearly 100% sure you get good genuine parts is to go for the 54 series mil spec ceramic package variants new old stock. In years of trying hundreds of these I have never found a faulty one yet.

 

[pjh]

Thanks for the replies. In my breadboard setup, I of course input from the 7493 and then I install LEDs on the outputs with the cathode to ground. In this configuration I will have an LED light up when the output goes high and turn off when the output goes low. So, with the 555 set up to approximately a one second pulse, I am watching the LED transition through the HIGH-DIM-LOW.

As for 'new' 74xx chips, I have also wondered if they are an accurate reproduction of the original or if they are a cheap knock-off.

 

[daver2]

You can identify the internal circuity of the 74LS30 device from the Texas Instruments data sheet.

A 74LS device has a VOL and VOH of 0.35V and 3.3V respectively. A 74LS device has a VIL and VIH of 0.8V and 2.0V respectively.

This means that anything below 0.8V will be seen as a '0' and everything above 2.0V will be seen as a '1'. So, you have to define what your 'DIM' voltage actually means.

If you look at the schematic for the 74LS30 device (in the TI data sheet) you will see that the pull-down to '0' side has a transistor, but the pull-up to '1' has a transistor in series with a resistor.

Dave

 

[Chuck(G)]

Piling onto Dave's comment, one thing that's sometimes forgotten is that the output of a TTL gate is intended as a current sink; that is, the gate is "on" when the output is low, not high. Expecting a TTL gate to perform as a current source (i.e. high voltage to low-voltage load) isn't playing by the rules.

 

[Hugo Holden]

In this configuration I will have an LED light up when the output goes high and turn off when the output goes low.

As mentioned in post 2 and Chuck mentioned in post 8, the TTL output stage is good at sinking current, not sourcing it. So if you are using an LED (and series resistor) as a load it should have its anode connected to the power supply and its cathode pointing toward the IC's output. The LED will be ON of course when the IC's output is low.

When you said in your initial post that you "watch" for clean transitions, I had assumed you were looking at the waveform on a scope ? But now I realize the "DIM" you were referring to is a low state of brightness of an LED ? If your LED is just connected from the TTL output to ground, with no current limiting resistor, the LED could upset the output stage in the IC due to its forward voltage drop. So I suggest move the LED and it must have a series resistor of at least 220R to 1k, and connect it between the +5v rail and the IC's output instead.

 

[pjh]

Thanks guys. I will connect the LEDs to the 5 volts and retest the questionable 74LS30s. It still presents an interesting question. With my incorrect setup, why is the 74LS30 the only chip so prone to the HIGH-DIM-LOW LED outputs while all the other 74LSxx chips don't seem to be as susceptible to that transition pattern. With my limited knowledge of electronic theory I have not been able to answer it.

 

[Chuck(G)]

There's a fair amount of quiescent current in a 74LSxx series IC--the output stages between most parts is the same; i.e., the output stage of a 74LS00 is identical to that of a 74LS30.

If the asymmetrical output characteristics bother you, you can always use a CMOS (74HC30 or 74HCT30), which go nearly from rail to rail on the output. HCT has a lower input "1" threshold than HC (usually 1/2 of Vcc) but is otherwise identical.

The key difference between CMOS and bipolar TTL logic is that one is voltage-controlled, while the other is current-controlled. That's why CMOS fanin is essentially limited by capacitance, versus input current for TTL.