• Please review our updated Terms and Rules here

Hardware 101

Z

ziloo

Veteran Member
Joined
Feb 7, 2006
Messages
994
Location
in the basement
Hello to all Hardware Masters,

Frequently in designing an SBC system, parts of a
CPU port are "open-collector". Would you please
explain about this arrangement and the use of it!

Much thanks...

ziloo
 
ziloo said:
Hello to all Hardware Masters,

Frequently in designing an SBC system, parts of a
CPU port are "open-collector". Would you please
explain about this arrangement and the use of it!
Click to expand...

<gross oversimpiification>

On most TTL or CMOS logic parts, there are two parts of the output of a logic gate--one that connects (or "pulls") the voltage level "down" or towards zero volts and the other that pulls the voltage level "high" or toward the supply voltage.

On most modern logic, these parts are generally transistors; think of two switches in series from the supply voltage to ground--you can close one or the other and the potential at the mid point where the switches are connected will go up or down. (as an aside, if you open both switches, the output will be neither high nor low, but "float"--this is the basis of so-called "tristate" logic in that the level can be high, low or floating).

Sometimes only half of this arrangement is implemented in transistors, and a resistor is substituted. Consider a light bulb in series with a switch. With the switch open, the level at the junction point of the two will the supply voltage (current flows through the bulb but not the switch); with the switch closed, the level will be that of the level on the other side of the switch and the lamp will be lit.

An open-collector breaks this arrangement in half and puts the switch in the IC and leaves it up to the circuit to supply the resistive load. This works well for driving longer cables--there is less "ringing" of the signal if the resistor connects the signal line to the supply voltage at the very end of the cable. Those terminators that you install in floppy drives are packaged bunches of resistors.

Another use is where several different circuits are used to drive a single line--you can have many switches, but only one lightbulb.

</gross oversimplification>

Does this help?
 
PLL, what is it?

PLL, what is it?

Thank you Chuck! I am looking up some stuff as examples,
and I am gonna come back again :geek:!!!

Now, why is it that everybody is all thumbs when it comes to
explaining about "Phase Lock Loop". I have read about it but
explanations are murky...:confused1:

ziloo
 
ziloo said:
Now, why is it that everybody is all thumbs when it comes to
explaining about "Phase Lock Loop". I have read about it but
explanations are murky...:confused1:
Click to expand...

I don't know--it's a pretty simple idea (and realized in vacuum tubes long before transistors). Maybe it's the idea of a "phase comparator" that gets confusing. :)
 
Re: PLL

Re: PLL

Hello Chuck,

1- How does it work?
2- What are the applications?
3- How does it work in a Tube?

Thanks

ziloo
 
ziloo said:
1- How does it work?
2- What are the applications?
3- How does it work in a Tube?
Click to expand...

Basically, a phase-locked loop uses a precision reference frequency, such as a crystal oscillator, and a variable-frequency oscillator. The reference frequency is divided down (think a chain of flip-flops, though other methods are used) and the phase of the divided down reference frequency is compared to the frequency of the varaiable oscillator. An error (or difference) signal is developed by a phase comparator and the error signal fed back to the variable oscillator, bringing it into agreement (i.e. operating at a sub-multiple of the reference). Think of how old-time piano tuners used to work--they'd strike a tuning fork (reference) and hit the key on the piano and listen for "beats" and then twist the tuning pin on the string accordingly (the error correction). All that was necessary is that the note being tuned bear some integral relationship to the reference, such as a fifth or third.

As far as applications go, consider an analogue TV signal. It's sent with a certain scanning frequency that's developed at the broadcast station. Your TV set has to track this signal to keep the picture synchronized. A PLL is used to lock onto the synchronizing pulses sent during the broadcast and you get to watch South Park reruns without the picture flipping. In this case, the "reference" signal is what's sent by the TV station, not a local crystal-controlled oscillator.

Suppose you're reading a double-density 5.25" floppy. The data clock for the bits recorded on the floppy are determined by the crystal oscillator on the FDC and the speed of the spindle motor. These are unlikely to be identical to those of the system used to read the floppy. However, by sampling the read bit stream, a PLL can be used to recreate the original data clock and determine the original timing of the bit cell transitions and you get to play with your Microsoft Flight Simulator.

Sometimes, the error signal from the comparator is what's wanted. Consider a PLL that's been synchronized to a FM broadcast radio signal. The PLL will track the average frequency of the broadcast over time, but the instantaneous error signal will reproduce the modulation that was applied to the carrier signal by the broadcaster.

Uses of PLLs are extremely diverse. It's a good thing to know about.

As far as implementing in a "tube", understand that just about anything that can be done with transistors can be done with tubes. A 1960 TV receiver has a PLL in it comprised of tubes. The principle stays the same.

I hope this makes sense.
 
Input Port

Input Port

They are telling me that in some microcontrollers such as 8051,
in order for the given port to function as an "Input Port", it
has to be placed in "High State" before inputting data on the bus,
otherwise-> bad fate!! Would you please explain that...

ziloo
 
Hi
Not sure what this has to do with PLLs
The 8051 didn't have a tristate bus. It was intended
to drive a mos input. It would pull down hard to ground
but after an initial hard pull high, it would have a soft
pullup that could be pulled down by the drive of another chip.
Dwight
 
Dwight Elvey said:
...Not sure what this has to do with PLLs
Click to expand...

Thank you Dwight for your response! I have had these questions in
my head for a long time, and I am patching things up!

All these pull ups and pull downs are making me dizzy :trampoline:
Would you please elaborate a bit more? How about Z80... is that
any different?

ziloo
 
Last edited:
AND Gates

AND Gates

In circuit design of vintage 8-bit computers, I come across a
combination of one or two AND gates along one single line (see
attachment). They are telling me that one purpose for such a
wiring is to shape-up a signal (at B) that could be a bit out of
shape (at A). Another use is to provide a slight amount of
"time dealy" for the signal at B. Any comment...specially about
the time delay?

ziloo
 

Attachments

  • Gate.jpg
    Gate.jpg
    4 KB · Views: 1
Re the Z80 and tristate, you may not need to use that on a simple board. The Z80 is always in control of the data and the address bus. But there is a pin called BUSRQ on the Z80 and that can be used by an external device to tell the Z80 to put all the lines into tristate. Then an external device could take over the busses and (say) move some data directly into the ram.

Re time delays and gates, say you had one signal path that went through 3 gates, (some OR and AND gates or whatever) and another path only went through one, then you might add two more gates to that second line so signals arrive at the same time. At Z80 speeds, and with nanosecond delays on modern chips it probably isn't all that important.

Then there is the signal delay due to the speed of light - 1 foot per nanosecond. That can add up on a big board. I noticed recently on a PC motherboard some of the traces take zig zag paths, probably so that all the traces of a bus are the same physical length. Or you could put in a gate to provide a known delay.
 
O-Scope Bandwidth

O-Scope Bandwidth

Hello Cosam,

According to wiki:

"Note however that a scope with 100MHz bandwidth cannot
measure a 100MHz digital signal with adequate definition."

I find this so very ruuuuude that I have invested 100s of dollars
in purchasing an O-Scope and I cann't get my money's worth :hammermon:.
Would you give more explanation about this :sneaky:?

ziloo :biggrin:
 
Dead Component,,,

Dead Component,,,

Hello Folks,

I have to check the deadness of capacitors, transistors,
diodes,.....and what not. Now, do I have to desolder/cut
one leg (two legs in case of a transistor) of the component
to test its performance, or there is/are other methods?

Thank you

ziloo
 
ziloo said:
"Note however that a scope with 100MHz bandwidth cannot measure a 100MHz digital signal with adequate definition."
...
Would you give more explanation about this?
Click to expand...
To be honest I don't remember the long and the short of it, so maybe someone else can elaborate. It has to do with signals being made up of not just a base frequency but several harmonics as well. If you loose the harmonics, you loose definition, e.g. square waves get rounded off.

I was given a good explanation once, which made great sense at the time. In such situations I tend to save brain capacity by just caching the short answer and not the entire reason ;-)
 
scope

scope

You should take the Wiki entry (and could we get a link to it, please ? ) with a grain of salt.
The scope's bandwidth figure usually refers the point at which the vertical amplifiers's gain has dropped by 3dB. What does this mean ? The frequency response typically begins to rolloff before the -3dB point, and continues beyond it. So technically the scope can display signals beyond it's rated bandwidth, but with an amplitude error.

So, when feeding a harmonic-rich signal, such as a square wave, into the scope, the edges of the waveform will become progressively rounded the closer you get to the frequency the -3dB point is at. It's exactly as if you're passing the signal through a low-pass filter, you still get a signal, just not what you started out with.

The trouble with pulse waveforms is that they, if you consider the cycle time, can be of relatively low frequency, but have very fast rise times (the transition from 10% to 90% amplitude is usually called rise time). Take the 8088 clock. If you look at the specs for a 8088 -1, you see that the clock runs at 10 Mhz, but the maximum (i.e. worst-case allowed) rise & fall times are 10 nS, which would be the period of a 100 Mhz sine wave. So even though it's fairly low-frequency, you've got a lot of harmonics, and to capture that definition is what you need the bandwidth for.

Usually you can make do with the x5 rule of thumb, use a scope that has a bandwidth five times that of the period you want to look at, so for an 8088, a 100 Mhz scope would do nicely.

If you want to look at the clock for a 60 Mhz Pentium, you'll still see a signal on the 100 Mhz scope, but the edges will be very rounded, and you might even have problems with the scope triggering stably, but you'll still see something. The clock for the P60 is only 60 Mhz, but the maximum rise & fall times are already 1.5 ns, which is way out of the league of the poor 100 Mhz scope.

If you just need to see if a signal is there, bandwidth isn't necessarily a concern, if you're doing deeper analysis such as if the pulses are shaped correctly, or ihow much jitter is present, then you need higher definition.

There's a brief overview on how to match bandwidth to application here:
http://cp.literature.agilent.com/litweb/pdf/5989-5733EN.pdf
or, if you're a Tektronix fan, here:
http://www.tek.com/Measurement/scopes/selection/pdf/55W_13768_2.pdf

patscc
 
in-circuit testing

in-circuit testing

Yes, you sometimes can, but you need to be careful & have a bit of an understanding how your circuit hangs together.
If you want to test the exact characteristics, you'll probably end up taking the component out of circuit.
If you're trying to do a basic functionality check, you can leave a lot of stuff in place.

Usually you start by trying to see if the path across a two-lead part is open or shorted. If it appears shorted, you look at the circuit to see if there's another path for current to flow that could be giving you that reading. Try reversing the leads, if it's a real short, that won't make much of a difference, but if there's something like a diode or a transistor junction in parallel to your component under test, then you might see a change. If it appears open, and it's a passive component, there's a good chance it's open, unless it's a capacitor. If it's a capacitor, especially a large one, it'll read a short while charging, and give confusing readings while it's discharging. Electrolytic caps may read fine (within their wildly varying actual versus rated capacitance) at the meter's voltage, but be leaky when operated near the rated voltage.
Semiconductor junctions (diodes, transistors, scr's, etc.) should have a high resistance in one direction , and a lowish one in another, depending on if your tester is using more than the 0.6 V or so (depending on junction material) required to turn to junction to conducting. If there's anykind of a snubber device integrated into the transistor, readings will be different.
Coils can be tricky, for similar reasons that capacitors are, usually you end up checking if their open, and then trying to 'ring' the circuit, if it's that type of circuit. Oh, and all the rules change if you're trying this on a powered circuit, or recently powered circuit.

Hmh. Reading the post thus far, I get the impression that it's a load of gibberish, and it is in a way, and isn't.
You really need a good idea of what the circuit does, how the components interact, and a certain amount of experience to get useful results with in-circuit testing. If you're still new to the whole field of electronics, most readings you take in-circuit will seem like gibberish. By all means, test stuff in-circuit to get a feel for the process and to see what's going on in the circuit, but for a final exact check, you'll usually end up taking the component out of circuit. With a bit of practice, though, in-circuit testing can help you narrow down the field to a few likely candidates for desoldering.

They do make various gizmos for in-circuit testing of various components, by the way, but again, they require a certain level of knowledge to get useful results out of.

patscc
 
Re "I have to check the deadness of capacitors, transistors,
diodes,.....and what not. Now, do I have to desolder/cut
one leg (two legs in case of a transistor) of the component
to test its performance, or there is/are other methods?"

Yes, probably. It depends on the complexity of the circuit, but say you were testing a diode and there were 20 other paths, some of which were due to other zapped components? It would be hard to isolate the one that is at fault.
There might be some simple circuits where you could test things in situ, eg if there was only one capacitor across a power supply. And with a signal injector, you can trace signals through digital circuits.

Or you can just do what the TV repair guys do:
"First, I replaced all the electrolytics. Then I started debugging the problem..."
 
From the wiki:

Note however that a scope with 100MHz bandwidth cannot measure a 100MHz digital signal with adequate definition
Click to expand...
(italics mine)

Yes, you can determine that a signal of 100MHz of some kind is there, but telling the difference between say, a square wave and a sine wave will be next to impossible, because waveforms can be viewed as consisting of combinations of higher-frequency components (see: Fourier analysis).

This is one key area where logic analyzer departs from an oscilloscope. In today's digital world, it's not unusual to see the two combined in one unit, but the logic analyzer is mostly concerned about the presence or absence of a signal, where an o-scope also gives you information about its form.
 
You must log in or register to reply here.
Back
Top