10Mhz squareish wave need to make TTL

Discussion in 'The Projects Forum' started by lordofentropy, Aug 23, 2012.

  1. lordofentropy

    Thread Starter Active Member

    Apr 10, 2010
    52
    0
    I have a 10Mhz square wave, whose Vp~.4 to.6 VDC, I have a pull up resistor on it to get the low end at ~0 VDC. I need to make this guy TTL or close to. I've tried some op amps, common-emitter transistor amp, however no success. Does anyone have a good, relative simple circuit to do the job that they can share with me, or have any suggestions? Thanks
     
  2. t_n_k

    AAC Fanatic!

    Mar 6, 2009
    5,448
    782
    How well does the output square wave have to track the 10MHz input? You could use a high speed comparator such as the LT1016 / LT1116 provided you can tolerate a delay/rise time of up to several (~10) nanosec. The layout and de-coupling would need to be well designed for 10MHz operation.
     
  3. ifixit

    Distinguished Member

    Nov 20, 2008
    638
    108
    Hi lordofentropy,

    What is the impedance of the 10Mhz signal source?

    An RS485 Receiver, or simular faster device, may do the job depending on the TTL output quality you need. These parts have a differental inputs with switching thresholds around 80 to 100 millivolts, which is what you need. A simple RC biasing network can convert a single-ended source to differental to give good noise rejection.

    Regards,
    Ifixit
     
  4. vk6zgo

    Active Member

    Jul 21, 2012
    677
    85
    I'm a bit confused,is the waveform 0.4V low,& 0.6V high,or is low 0V & the high varying between 0.4V & 0.6V?

    In either case, it is a quite low level signal, so if you use a common emitter transistor amplifier,you have to either AC couple it into the base & separately bias the transistor,or pull the low up to around +0.7v,or it won't amplify.

    If you can increase the amplitude to around 5V p-p,you then will have to restore the low to 0V,which can be done with a "DC restorer" type of circuit.
    Another suggestion is to have a look for video amplifier circuits,as they have similar problems,particularly those in video cameras.

    It is very challenging to design things from scratch,but there is almost certainly a ready made circuit out there which can do the job with a few modifications.
     
  5. lordofentropy

    Thread Starter Active Member

    Apr 10, 2010
    52
    0
    I do not know the exact impedance of the signal, however the circuit is laid out for 50 ohm impedance matching, so...50 ohm? Whats going on here is that my initial signal is a 97Mhz sine wave, that I have divided by 8. So actually 12Mhz. Using a pull up resistor, I have the base of the waveform right about 0 VDC. Then of course (now 12Mhz) the square peaks at around .4 to .6 VDC.

    I'll pull up the signal to start at .7V and try it again with common-emitter transistor, I had +- 6 VDC bias on the transistor's collector/emitter. Also had cap in series with signal input.
     
  6. lordofentropy

    Thread Starter Active Member

    Apr 10, 2010
    52
    0
    What about using a mosfet switch?
     
  7. ifixit

    Distinguished Member

    Nov 20, 2008
    638
    108
    Hi lordofentropy,

    To do a reliable discrete circuit with mosfets, fets, or transistors for this application would be very difficult. It requires way too much expertise and time to design and test properly. Why should you when there are ready made solutions available.

    The LT1116, suggested by t_n_k, seems to work, but the rise and fall times are not great. Are they good enough? I believe a fast RS485 type receiver would be a better choice if configured as shown below.

    Regards,
    Ifixit
     
  8. lordofentropy

    Thread Starter Active Member

    Apr 10, 2010
    52
    0
    The times are fine, I just don't have one laying around at the moment. I do have AD8611 and ADCMP601. I have not compared the datasheets in detail against in the LT1016/1116; but glancing through they seem similar enough so I am going to try what I have. Any suggestions on the AD8611 and/or ADCMP601?
     
  9. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    How did you divide the signal by 8?
    Why do you have such a low voltage swing?
     
  10. lordofentropy

    Thread Starter Active Member

    Apr 10, 2010
    52
    0
    Ron, at first tried with ECL positive edge triggered flip flops, was going well on board until I reached a certain frequency say around 50/60Mhz after that circuit wouldn't perform...probably due to layout; did not spec out a board to be made. So in the end I got some of these guys http://www.hittite.com/content/documents/data_sheet/hmc394lp4.pdf

    I have a divide by 8 and divide by 10 of the same 97Mhz in, the divide by 8 needs to be TTL and w/ that HMC394 (800mV swing). The other I don't need TTL rather I need to take the DC part of it out and filter it some to make it more sinusoidal. They are all off of the same clock, so it can sync up with various external electronics on the same clock.
     
  11. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    It looks like your part has differential outputs. Check out this differential input PECL-TTL converter. It has a wide common-mode range, which should easily accommodate your outputs.

    EDIT: This part is not stocked by Digikey, so is probably hard or impossible to buy in small quantities.
     
    Last edited: Aug 24, 2012
  12. lordofentropy

    Thread Starter Active Member

    Apr 10, 2010
    52
    0
    Tried the circuit that tnk posted w/ the AD8611, and did not work out. In fact, did not amplify at all, seem to degrade it some. I'll order some LT1116 Monday, but being that its similar to the AD8611 I don't think its going to work. I am testing this part of the circuit on a breadboard. Also, in the simulations you all ran, the input waveform isn't a sinusoidal, its a square wave (not a perfect one); none the less it should still pick up on the edges.
     
  13. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    I just realized that the part I recommended is non-stocked by Digikey.
    Try LT1711. It has rail-to-rail IO. Connect the true and complement outputs from your divider directly to the inputs of the comparator.
    LT1116 will have the same problem as AD8611, because the input range does not go near enough to the positive rail.
     
  14. lordofentropy

    Thread Starter Active Member

    Apr 10, 2010
    52
    0
    This seems similar to the ADCMP601, I have not tried it with connecting both the true and complement outputs of divider to the +/- inputs of comparator yet. Configured similar to the other comparator w/ just the one input it did not work. Should I ground V- or put -5V on it b/c of differential input? Assuming to ground the latch as well.

    Oh, and thank you all for your input/help. I am stuck on this and have been poring through books/internet to figure this thing out. It's become personal to me now =\.
     
  15. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    The ADCMP601 should also work.
    On the 1711, connect V- and the latch input to ground. High speed comparators are prone to oscillations. You need very good power supply decoupling and a good ground plane.
     
  16. lordofentropy

    Thread Starter Active Member

    Apr 10, 2010
    52
    0
    Your pretty much talking about doing a grounded coplanar waveguide layout.
    As far as the bypass cap's, I always wondered about when someone says very good decoupling. I just always put a few low value ceramic caps in parallel w/ each other. But what does someone exactly mean by that, i.e. how many and range of the caps? Are we talking just one at Vcc...1000pF?...another one on the board where power goes in say 10uF; whats very good enough, whats the limit here (over doing the bypass cap's).
     
  17. Ron H

    AAC Fanatic!

    Apr 14, 2005
    7,050
    657
    Not necessarily. It depends on trace lengths and signal transition times. Even if you do need controlled impedance traces, I prefer microstrip to coplanar, because it doesn't slice up the ground plane. Coplanar is good for tapering a line to a pin while maintaining the impedance.
    A good ground plane provides a low-impedance return path for high frequency signals and power supply transients.
    You can generally get by with a 100nF ceramic cap (SMD chip caps are best) as close to each IC power pin as possible, with the other end connecting to the ground plane, also with the shortest possible trace length. A via directly to the ground plane is best.
     
Loading...