Step-by-Step Assembly Guide

1. Add the short stuff

Add all the resistors and diodes.

2. Add the DIP-8 sockets

Add sockets for IC1, IC2, IC5, and IC6. Leave the chips out for now, even if you’re going to solder them to the board.

3. Add the TO-220 parts

Add the LT1206 and the regulator, and put heat sinks on them if needed.

You will need to push the parts down in their holes as far as they’ll go if you’re using the Hammond 1455L12 case, to avoid having the part hit the top of the case.

4. Add the capacitors

Add all the caps. Be sure to get the polarity right on C1, C3, C6, C10, C12, C16 and C17.

5. Add the LEDs

The amp power LED can be soldered directly to the board, with enough lead length left to allow you to bend it into position to fit into a front panel mounting hole. I recommend a ‘Z’ kind of shape with the board forming the lower part of the ‘Z’. This gives you flexibility to bend the LED around to meet your mounting hole.

The charge indicator LED will need to be soldered to hookup wires long enough to reach the front panel, if you want to be able to monitor the charging; otherwise you can just solder the LED straight to the board. The polarity isn’t marked on the board; the negative lead is towards the rear.

6. Do the wiring

Solder the battery holder wires to the board. If you’re gluing two 4×AAA holders back-to-back, you’ll have to tie the center pair to each other directly; the board has no provision for this.

If you want the 40/60 dB gain selection feature, add S1. If you’re using a front-panel switch, you can either run the wires to the switch now, or just add the wires and strip the ends so you can touch them together. If you don’t want the feature, leave S1 open if you want 40 dB gain, or add a jumper if you want 60 dB gain.

Add the power switch. You need wires from the S2A, S2B, S3A and S3B pads on the board to the switch. The wiring instructions are on the second page of the schematic.

Add the input and output jack wires. You can strip the ends and leave them bare for the moment, or you can solder them to the jacks if you want. It depends on whether clipping onto the wires or using the jacks will be easier for you during testing.

7. Test the amplifier power supply

If you’re using two 4× battery holders, I recommend using double-sided foam tape to bind them together. This pads them out to be tall enough that they fit snugly within the case so they don’t rattle; the foam compresses a little bit when you shove the batteries inside the case.

Populate the battery holder. Measure DC volts from V- to S3B while toggling the power switch. You should get the battery voltage when PLED comes on, and nothing in the other position. Ensure that the polarity is correct. If it’s incorrect, IC3 is probably toast now.

Throw the power switch so PLED turns off. Install IC6 and toggle the power switch. Observe that PLED comes on. Measure DC volts from pin 1 (black lead) of IC6 to pin 3 (red); you should get approximately 5 V. Move the red lead to pin 2; you should get very close to the same value only negative. If so, the rail splitter is working. If you want to be paranoid, do the same check from IC6 pin 1 to the V+ and V- pins of IC1-3.

8. Test the battery charger

With PLED on, install IC5.

Plug in the wall supply and measure DC volts from S2B to V-. You should read nothing on the meter.

Toggle the power switch; PLED should go out, you should see the wall supply’s voltage on the meter, and CLED should begin flashing. If CLED comes on solid, the voltage at the midpoint of R16 and R17 is outside the normal 1 to 2 V range. If this value is below 1 V, the battery may be badly depleted, and it should be brought within normal range in a minute or so by the trickle charging path; CLED will then start blinking. If the voltage stays below 1 V, one or more cells is bad, in all likelihood. If the voltage is above 2 V, the resistor values are the wrong value for the battery pack; see the MC3334x datasheet for details on picking proper values.

If the above tests pass, measure DC volts across R14. You should get 1.25 V, which will fluctuate down momentarily about every second, in time with each CLED flash. Let it run for a minute or so. R14 and the regulator should be getting warm now.

If the batteries are new, or haven’t been on a charger for more than a month, let them charge for at least 15 minutes before proceding.

9. Test the amplifier’s basic functionality

Unplug the wall power supply. Toggle the power switch so that both CLED and PLED are off.

Install IC1 and toggle the power switch. PLED should come on and the chip shouldn’t get hot. Apply a 1-10 mV signal to the amplfier’s input and see that it comes out about 10× as high at the output of IC1.

Toggle the power, wait for PLED to go out, install IC2, turn the power back on, and measure the signal at the output of IC2. The normal range is 87 to 113× the voltage at the amplifier’s input.

Measure the signal at the output of the amplifier with S1 open: it should be very close to the value you measured in the previous step. Then close S1 and remeasure. At this point, you just need to see that the voltage goes way up. Ideally, it should be 10× higher, but the value could be quite a bit lower without calling into question whether the amp is working properly. You need to complete the next step before worrying about absolute values. I would only worry if you read something well over 10× the previous measurement.

10. Trim the amp gain

Up to this point, the amp’s accuracy has been limited by the components’ inherent accuracies. If your goal in making the LNMP is to make qualitative or ratiometric distinctions — as was my goal in designing it — then you should do this step only as a best effort attempt at improving the amplifier’s accuracy. If absolute accuracy is more important to you than it was to me, you will have to go to some effort to make that happen. All else being equal, you must use a more accurate instrument to measure a less accurate one. To avoid spending more on the LNMP than you could on a pre-made, calibrated measurement amplifier, you will have to make all else not equal, by use of clever calibration and measurement strategies.

Another concern is that if you’ve added a gain selector switch, trimming the amp for perfect gain on one gain setting won’t give perfect gain on the other setting with the stock LNMP design. This did not concern me in designing the LNMP because the only reason I added a 40 dB setting is for those rare times that I have a signal over 1 mV and the amplifier is clipping it. Once the signal is over 1 mV, I can usually use my other test equipment without the LNMP’s aid, so I don’t have a good reason to worry about this setting’s accuracy. If that’s not your situation, you have two options:

• Trim the gain to split the error difference.

• Use an adjustable resistor in R11 instead of a fixed one so that you can trim both stages for accuracy. This will be an interactive process: you’ll have to trim them each repeatedly, approaching the proper balance carefully.

From here on out, I’m going to assume that you also will be using the 60 dB setting primarily, and are trimming for best accuracy on that setting. If not, adjust the advice below as required.

The procedure to trim the amp’s gain is conceptually simple: run a known test signal through it, and adjust R6 until you get a value 1000× as high at the output of the amplifier. There are several interacting subtleties that you need to be aware of in actually doing the adjustment:

First, the LNMP’s gain accuracy can only be as accurate as your measurement of the test signal. I believe the best way to get high accuracy at the lowest cost is to take advantage of the high sales volumes in consumer electronics: the price for a given level of accuracy will be much lower here than in proper test equipment. Choose your highest quality music player, make it play a 0 dB 1 kHz tone, and measure that voltage. (See this section of my audio troubleshooting article if you don’t know how to get test tones from your chosen source.) Now you need to divide that down to 1 mV, or thereabouts. If the 0 dB tone is 2 V, then 1 mV is a factor of 2000 down from that, or -66 dB. (Read this article if you are not clear on how I got that value.) Change the test tone’s level appropriately, run that through the amp, and adjust the gain until the output signal is 1000× the input value.

Second, you must be wary of the amp’s clipping point. At minimum battery voltage, it’s just over 1 V, so the signal must be 1 mV or less to avoid clipping. Even with a full battery, you can’t use a signal greatly higher than that. But at the same time, you want to use a signal as large as possible to avoid losing too much of the signal in the noise floor. If you have a scope, you might experiment with this, to find the exact clipping point. Then give yourself some margin under this, because the battery’s voltage will be falling as you continue the calibration process. Beware that true clipping begins before you can see it on a scope; you’ve typically added 1% or more distortion to the signal by the time clipping is visible.

Third, the LNMP’s input impedance is 100 Ω, so unless your source’s output impedance is under 1 Ω, the voltage divider you create will be significant enough to throw off the adjustment if you ignore it. If you know the source’s output impedance accurately, one way you can cope is to simply calculate the divider value and take that into account when calculating the expected gain. If not, I would suggest constructing a simple inverting op-amp stage between the generator and the LNMP. This is just an op-amp, a single-voltage power supply, and a few passives. Going with an inverter instead of a buffer will keep the circuit simpler, which will help because phase doesn’t matter for this test. Make it AC coupled so you don’t need to worry about DC offsets. See Appendix A in Op Amps for Everyone for circuit ideas. If you have a spare op-amp for the LNMP, that will work beautifully. If not, even a TL082 from Radio Shack will do the job here.

Fourth, you can’t expect to get high accuracy if you use a crappy meter or a cheap signal source or an MP3-compressed test tone. Your ability to make meaningful comparisons is limited by the quality of your testing chain. Garbage in, garbage out. One source of error you might not be aware of is that a test tone mastered at -60 dB will likely come out of the player significantly higher than -60 dB, owing to noise and distortion within the player. This gets increasingly worse as the voltage drops, and more so on cheaper players. This is why it’s important to use a high-quality player for this, such as one made for the newer 24-bit media, like a SACD player.

11. Finish up

Finish up any remaining casework, close it up, and you’re done!