Announcement

Collapse
No announcement yet.

Full m/lab SD2000 schematics

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • liekwang
    replied
    what is the BAS16 diode replacement can I use diode 4148

    Leave a comment:


  • Carl-NC
    replied
    Originally posted by GoldenKiwi View Post
    I'm still scratching my head over this input stage.
    When I first looked at it, I assumed it was a composite of a single-ended differential BJT input providing gain at very low noise, followed by an op-amp as second stage of gain. Having built a Hammerhead years ago I recalled reading something to effect that splitting the gain (default x1000) into two stages provides better settling through maintaining the gain bandwidth product - and assumed this was essentially the same principle.

    I assumed the gain of the LM394 was essentially 1350/560, while not understanding the current source arrangement, multiplied by the opamp gain, which I wasn't sure how to calculate in this arrangement(positive feedback?). This is clearly not the case!

    What am I missing? I now guess the LM394 is not singled ended and the feedback loop is including the LM394. Is this practically including the BJT LTP as the input of the opamp?

    Apologies to ask what are probably basic questions!
    Q13 and R112 are not directly part of the gain. They are the current source for the diffpair; the collector current of Q13 is (5 - 0.75)/560 = 7.59mA. Let's call it 7.5mA. This current gets split in half in the LM394 diffpair so each leg sees 3.75mA, and that is the current through the collector resistors, which is 1.35k each side. Thus the DC outputs are sitting at about 0V.

    The gain of the diffpair is "everything in the collector divided by everything in the emitter." The collector side is easy, it's just 1.35k. The emitter side has no physical resistors, just the impedance of the emitter junction itself. That is called "re" and is VT/I where VT is the thermal voltage kT/q and I is the DC current. We know I -- it is 3.75mA -- and VT is 26mV at room temp. Therefore re = 26mV/3.75mA = 6.9 ohms. So the overall gain of the first stage is 1350/6.9 = 195. It is fully differential.

    I expect the diffpair was added to the NE5534 to reduce noise. The LM394 has about half the input-referred noise of the 5534 (3.5nV/rtHz). With today's opamps (down to 0.9nV/rtHz) it doesn't make much sense, but there are other discrete front-ends that can beat the noise down to 0.5nV/rtHz, so it's still a valid technique.

    Leave a comment:


  • GoldenKiwi
    replied
    I'm still scratching my head over this input stage.
    When I first looked at it, I assumed it was a composite of a single-ended differential BJT input providing gain at very low noise, followed by an op-amp as second stage of gain. Having built a Hammerhead years ago I recalled reading something to effect that splitting the gain (default x1000) into two stages provides better settling through maintaining the gain bandwidth product - and assumed this was essentially the same principle.

    I assumed the gain of the LM394 was essentially 1350/560, while not understanding the current source arrangement, multiplied by the opamp gain, which I wasn't sure how to calculate in this arrangement(positive feedback?). This is clearly not the case!

    What am I missing? I now guess the LM394 is not singled ended and the feedback loop is including the LM394. Is this practically including the BJT LTP as the input of the opamp?

    Apologies to ask what are probably basic questions!

    Leave a comment:


  • GoldenKiwi
    replied
    Many thanks waltr, green, ZED! VERY much appreciated.

    I was pleased to hear the circuit isn't the easiest to understand!

    Ah! I assumed the 1n+500r would likely roll off the HF but didn't know its called a snubber or how to calculate.

    Many thanks for the spice model. I've installed LTSpice and so far managed to observe the gain, perform the freq sweep, and just now starting tinkering and learning. This has pushed me to use Spice for the first time.

    Thanks for the links. Ill have a read. I have Doug Self's Small Signal Audio Design book, which is also good.

    I have purchased a tidy SD2000 cheaply in untested state. First step will be making sure it works. Step 2 id like to try a few updates and mods. This front-end architecture seems to offer advantages over the later opamp input, so like others have I probably will refresh a few things like trying the SSM2212/MAT12.

    Thanks for your help.

    Leave a comment:


  • waltr
    replied
    Good paper on audio amps.

    That SD2000 works so the 200kHz bandwidth seems ok.

    We do need Fast >1MHz GBW op-amp to obtain good linearity with a 200kHz cut-off.

    Leave a comment:


  • ZED
    replied
    Thanks waltr ! and so do you think the circuit is fast enough for a P.I. preamp ? is 200kHz fine or is say 5mHz better ? i have read a variety of opinions.


    Heres a good read on various parts of a preamp circuit.

    Leave a comment:


  • waltr
    replied
    Thanks for sim model.
    I did this but got wrong connection (gnd instead of -5V) to current source.
    Ran a freq sweep and Gain is 33dB from 0 to 20kHz. That is a gain of 44.
    Close enough to your result.

    The 1n & 500r rolls off the gain by -6dB at 240kHz.

    Leave a comment:


  • ZED
    replied
    Yes green yer right mate the gain is around 47,thats how i would calculate it too.

    GoldenKiwi heres a sim of the cct your interested in,just unzip it,hope it helps,with my basic understanding of electronics this preamp cct is way out of my league.
    This preamp is only found in the SD2000,the later machines just use a 797 chip with receive fets.
    The bias on the constant current circuit can be done in any number of ways,generally you want to minimize drift with temperature and so using the temperature coefficient of a diode is a big help but
    a pair of resistors in this instance is as basic as it gets.
    Long tail pair preamps like this one are great for minimizing DC offset on the output when used with matching transistors.
    The four resistors on the collectors i imagine are to improve the tolerances between them,the tighter the match between the collector resistors the better the DC offset is.
    Plus waltr is spot on with regard to the 1n cap and 500r resistor.

    Also the spice model for the LM394 isn't available so the SSM2210 was the next best thing.Dont remember where i got the model from though but thanks all the same.
    Attached Files

    Leave a comment:


  • green
    replied
    Is the gain (1000/22)+1=46.5?

    Leave a comment:


  • waltr
    replied
    That is a very interesting circuit and not the easiest to understand.

    The 1nF-150 Ohm are in series. At DC to low frequencies the 1nF cap is high Z (open) but at higher frequencies the cap conducts putting the 150 Ohm across the NE5534 inputs. I call this a 'snubber' as it attenuates high frequencies and Fast impuses (pulse edges).

    The LM394 & NE5534 is a basic instrument amp. Check this App Note:


    I also do not know how to calculate the exact gain.

    Leave a comment:


  • GoldenKiwi
    replied
    Originally posted by Kev View Post
    For the record, most of the answers are all over on Doug's forum.
    I had to test it out for myself, I tried manipulating the 33k, and found that it needs to be between 30k and 40k so 33k is perhaps optimum.
    The gain resistors also appear to be optimum.
    That only leaves the emitter tail current. Data sheets suggests a current between 3 to 4mA gives the lowest noise figure.
    So this is probably what unregistered was on about. The bias on the current regulator transistor is beyond me and may be better with a simple diode stabilizer.
    Also the compensation resistor maybe better around 220 ohms, especially if the tail current is trimmed fro lowest noise, but I've not tried that yet.
    I'll leave it there.

    Cheeeers
    Hi, I'm looking at the SD2000 LM394 into NE5534 based front end and am trying to understand it. I've searched and seen this discussion but I don't recall any specific answers. I would love some help if possible to try and understand.

    What is the function of the 150R and 1NF between the legs of the LM394 LTP?
    I'm am unsure how to calculate the emitter tail current with the current source arrangement, nor the resulting gain of this first stage. Can anyone suggest a specific link of similar arrangement and calculations or please describe the calculation?
    I am trying to understand the collective gain of this composite first stage (LM394+NE5534). Elsewhere the gain was suggested to be 6000-8000. https://www.geotech1.com/forums/show...highlight=6000

    I'm trying not to be lazy...I've ordered a copy of "The Art of Electronics" to try and help me.

    Any help would be very much appreciated!

    Leave a comment:


  • Bayu
    replied
    G.s how many pcb size and width please

    Leave a comment:


  • Andrei
    replied
    It is interesting to see how minelab did it, but on the Internet there are few photo boards sd2000

    Leave a comment:


  • Andrei
    replied
    Originally posted by Mechanic View Post
    I'm not Zed, but I'll answer that..

    The sd2000 board is only 2 layer. They were all 2 layer up to the gpx, the GPX uses 4 layer due to the layout density.

    EMI comes in through the coil.


    Cheers Mick
    Ок, thanks!
    my Board turned out to be noisy, I decided to make a different trace, use the top layer of the same for the ground polygon and ground the smd components. and there was a question, whether it is necessary to divide polygons of the earth of different modules: transmit cct, logic cct, receive

    Leave a comment:


  • Anis
    replied
    much appreciated for all your support!
    Thank you all

    Leave a comment:

Working...
X