Line drivers

[keithjeff]

Greetings,
I'm looking for a line driver that will drive about 1 metre of coaxial cable with TTL level signals at frequencies up to 10MHz. This means that it must be capable of significant current sourcing. I've tried the usual system of parallel 7404 units. Unfortunately, low jitter and non inversion is required.......
The unit I was looking at was the 7417 type; however it's only available in TSSOP and is open collector.
Thanks all,
Keith

[Manul]

How low jitter? If you want low jitter you also need well regulated supply voltage and good bypassing. As you said, currents will be high.

[keithjeff]

How low jitter? If you want low jitter you also need well regulated supply voltage and good bypassing. As you said, currents will be high.

I can live with up to about 10nS jitter. Good bypassing, yep, although the bypass caps will supply some of the current required to feed the line. Well regulated, yes. In this circuit there are two PSUs. One to drive a Xtal oven at 10MHz, the other to power a PicDIV to give a 1Hz output with varying duty cycle and the line drivers. Lots of electrolytics and SMT caps about the place. The 10Mhz doesn't need to be hugely accurate. A GSPDO would be overkill.
You'd think this would be simple. Ah well c'est la vie.
Keith

[moffy]

I built this for signaling over CAT5 it might be of interest: https://www.eevblog.com/forum/projects/cat5-fast-cheap-digital/msg6305732/#msg6305732

[voltsandjolts]

For ideas, try searching for 10MHz distribution amplifiers, e.g.
http://www.wriley.com/An%20Octal%2010%20MHz%20Distribution%20Amplifier.pdf

[keithjeff]

I built this for signaling over CAT5 it might be of interest: https://www.eevblog.com/forum/projects/cat5-fast-cheap-digital/msg6305732/#msg6305732

It is of interest. Thank you very much....
Keith

[MarkF]

Instead of chips with a standard TTL output drive, look at chips with higher bus driving capabilities.

I've used 74244 and 74245 line drivers with higher (typical 20mA) source and sink outputs in
the standard or H series.

The basic logic chip only can source 0.8mA and sink 16mA.

You also might look at the SN74LVC1G17 buffer chip which has 32mA outputs.

[mtwieg]

What is the load impedance on the receiver side? Is there a broadband 50ohm load termination, or just the input capacitance of the receiver devices?

What sort of rise/fall times do you need at the receiver?

Depending on the answers, you might not need any fancy driver, just a basic logic gate with series termination.

[PCB.Wiz]

Greetings,
I'm looking for a line driver that will drive about 1 metre of coaxial cable with TTL level signals at frequencies up to 10MHz. This means that it must be capable of significant current sourcing. I've tried the usual system of parallel 7404 units. Unfortunately, low jitter and non inversion is required.......
...
I can live with up to about 10nS jitter.

10ns of jitter is a lot, so not a problem.

What did you try exactly? Did you use 74LVC04?

10MHz and 1m is also not difficult, especially if you only take off the far end. (so the near end step is unimportant)

Then you just need appx source matching, so something like a 74LVC14 with 1 gate as inverter driving 5 parallel gates driving a resistance of ~ (Zo-5) ohms
would be a reasonable starting point.

examples : https://www.ecircuitcenter.com/Circuits/tline1/tline1.htm


Good decoupling would be needed as you have short ~50mA pulses (~10ns) into the cable.

[Andree Henkel]

you need 50mA drive current: 5V source, source termination 50Ohm, receiver termination 50Ohm; 100Ohms in total, 50mA current
gives in theorie 2.5V High level at termination; TTL level means you need 2V High level at Termination; so there is margin for some cable losses, tolerances of source/load resistors, 5V supply tolerances and fact that driver output is less than supply

a triple lvc buffer at 5V: SN74LVC3G17; 130Ohm at each output, connected together after the 3 resistors will work just fine
this buffer is push pull, not open collector

some more drive power you get with an 74ABT245 at 5V, four, or even all eight channels and suitable current share resistors

[jonpaul]

check RS-485 line drivers.

exactly whats needed

j

[David Hess]

For a point-to-point link, double termination is *not* required.

For instance when only a series termination is used, the voltage step on the near side is half of the drive voltage, divided between the series termination resistance and the transmission line impedance.  When the half voltage reaches the open unterminated endpoint, the voltage doubles back to the original drive voltage, and when that reflection gets back to the driver, it is absorbed in the series termination.

For a parallel termination, the driver drives the transmission line to the full voltage, which then would double at the end of the transmission line but the impedance into the parallel termination halves that back to the original voltage with no reflection.  This is the method that TTL would normally use, but if the transmission line impedance is 50 ohms, then a special TTL driver like the 74140 or 74S140 was required to drive that low an impedance and product TTL levels at the receiver.  TTL usually worked with transmission lines of 100 to 150 ohms so this was not a problem.

That leaves some options.  For highest performance, I like hacking up a custom driver using a couple of transistors to drive the transmission line with a current that has a defined voltage compliance producing the correct logic level and no aberrations, but that is not required here.  (2) An HC or AC CMOS gate will drive a 50 ohm transmission line with a series termination of about 25 ohms to produce a 5 volt signal.  Put several gates in parallel for better reliability (1) and increase the series resistance for the proper termination impedance.  Double termination under this condition will produce a 2.5 volt high, which is technically high enough for TTL, but barely.

2.5 volts into a 50 ohm parallel termination will require 50 milliamps while 5 x 7404 hex inverters in parallel (the extra is used to invert the signal making the driver non-inverting) will only provide 25 milliamps, so a 7404 with parallel termination will not work, at least with a 50 ohm transmission line.  A series termination will work though, but can you count on the receiver not being terminated?

(1) Divide the current up between parallel gates because CMOS has pin limitations on current, so pulling full current from one gate may be unreliable.

(2) If I did not use some gates in parallel, I would do it with a two transistor PNP current mirror that has maybe 10x current gain.  This inverts, so drive it with an inverting gate or add a third NPN transistor inverter.  The gate or open collector output drives the input to the PNP current mirror with a resistor which sets the input current, which is amplified by the current mirror, which pulls the transmission line up to about 4.5 volts.  Parallel terminate the driver side and it should not matter if the other side is terminated or not, but set the driver output current to assume both sides are parallel terminated.