Part Selection Guide
Some of the parts below are optional. For most of these parts, you simply leave them out if you don’t want them. If a jumper is required, it will be mentioned.
Resistors
Most of the resistors need to be 1% metal film resistors in order to meet the noise and accuracy specs. The only exceptions are R6 (a trim pot) and R14 (a power resistor).
All of the resistors should be ¼ W, except for R14 which should be 1 or 2 W. Although a ¼ W resistor will survive in R14 in some configurations, it will get awfully hot, so it will drift and maybe die young.
If you want to use the Vishay Dale RN/CMF series resistors, the RN55s are ¼ W at the temperatures you should see in this circuit. They’re specified for lower wattage because their temperature range is so high.
R1, R2, R3, R4, R5
These are gain-setting resistors for the first two stages. If you want to use different values, beware that the amplifier’s noise floor will rise if you raise their values.
R6
This is trim pot to adjust the gain of the circuit to account for gain-setting resistor tolerances. It makes a parallel/series combination with R4 and R5 to give a gain variation of about 9 to 11× in the second stage, which is sufficient to cover tolerance errors for the worst-case combinations of 1% resistors. Therefore, it is not necessary to do any resistor matching in this circuit.
R7, R8
These resistors adjust the behavior of the high-pass filter that’s part of the third gain stage.
If you want to pick other values, I recommend using Analog Devices’ Filter Wizard. These resistor values should be kept fairly low in order to maintain good noise performance. Their values must go up if you lower the values of C7 and C8, and those caps are the largest caps suitable for filters you’re likely to find, so these values are pretty well tweaked-out already.
R9
This resistor sets IC3’s bandwidth. It should be about 5 MHz with the default value, all things considered. See the LT1206 datasheet for details.
This resistor’s default value assumes that you’ve added C4, and that its value is above 100 pF or so. If you’re using a lower C4 value (or no C4 at all), you may need to lower this resistor value to increase IC3’s bandwidth; this comes at the cost of higher quiescent current, and you will probably also need to adjust R10, R11, and R12 to maintain gain flatness. See the LT1206 datasheet for details.
R10, R11
These are gain-setting resistors for IC3. Because IC3 is a current-feedback op-amp, the value of these resistors is fairly critical. See the LT1206 datasheet for details.
There are other resistor values that would get closer to an ideal 10× gain, but they would sacrifice gain flatness. This pair is the closest I was able to come without sacrificing one or the other too much.
R12
This resistor serves two purposes: as part of an RC filter with C12, and to load IC3 with a known value in order to maintain gain flatness. If you want to select a different value, you should study the LT1206 datasheet first.
R13
This is the battery charger’s trickle charge current setting resistor:
R13 = (Vs - 1.4) / It
where:
Vs = Wall supply voltage
It = Desired trickle charge current
The “1.4” represents the D1 and D3 diode voltage drops.
The trickle charge current shouldn’t be more than 0.1C for NiMH cells. (E.g. 70 mA for 700 mAh cells.) Lower is okay, especially if you mainly want the trickle charger to maintain the battery’s charge.
R14
This is the battery charger’s fast-charge current setting resistor:
R14 = 1.25 / If
where:
If = Desired fast charge current
“1.25” is the voltage drop from the regulator’s OUT to ADJ pins. Although the voltage drop across R14 isn’t precisely 1.25 V (it forms a voltage divider with R15) it’s close enough to the real value when R15 is much larger than the value of R14.
The fast charge current shouldn’t be more than 1.0C for NiMH cells. (E.g. 700 mA for 700 mAh cells.) Lower currents will keep the generated heat down, prolong cell life, and allow you to use a smaller power supply. The default value gives about 260 mA, which should give a full charge in under 4 hours. I personally wouldn’t go much higher than 500 mA due to the physical size of R14 and the available space for the regulator’s heat sink.
The best resistors for noise purposes are wirewound types. Next best would be a large metal film type, but those are only rarely available over ½ W. Metal oxide should also be reasonable here. I wouldn’t use any carbon type here, since it’s directly connected to the V+ rail.
R15
This resistor limits current from the ADJ pin to just high enough to keep the regulator operating. You pick its value using the same formula as for R14. Because you only need 5 to 10 mA through R15, ¼ W resistors are just fine here. See your regulator’s datasheet for the details.
R16, R17
These resistors divide the battery voltage down for the charge controller’s battery voltage sense circuit.
The default resistor values are correct for an 8-cell NiMH battery pack. If for some reason you’re using a different size pack, see this section in the PPA battery board project’s parts selection guide. (This circuit is derived from that one, and its documentation goes into more detail on picking resistor values.)
R18
This is the charge indicator LED’s current limiting resistor.
R18 = (V+ - Vf) / If
where:
V+ is the wall supply voltage
Vf is the LED’s forward voltage drop
If is the desired current through the LED
1 mA gives enough brightness for a power indicator with most LEDs, but some may require a bit more. You don’t want it to be too bright, or it’s annoying. Typical values for RLED are 1 kΩ to 10 kΩ, depending on the power supply voltage and the LED being used.
R19, R20, R21
These parts are for setting the backup fast-charge termination mode. Normally the charge controller can terminate fast-charging correctly by sensing the pack’s voltage, but there needs to be a backup method in case something goes wrong with the cells. For instance, when a cell starts getting weak, the stress of fast-charging can make it overheat.
If you don’t want to think too much about this and you’re using the default value for R14, I recommend putting a jumper across R20 and leaving the other two open. If you want to do something different, see this section in the PPA battery board project’s parts selection guide.
This circuit also supports temperature-based backup fast-charge termination; this is covered in the MC3334x datasheet.
RLED
This is the current limiting resistor for the amplifier’s power indicator LED. You pick its value the same way as you do for R18.
Capacitors
The high-value caps (10 µF and over) in this circuit are tricky because they all need to be high quality. Heed the advice below carefully, as cheaping out on these caps will significantly hurt the performance of this amplifier.
Most of the smaller-value caps can be generic types. The exception is C4, which is a filter cap so it needs to have a low value tolerance.
C1 (electrolytic)
This forms an RC high-pass filter with R1. Its purpose is to strip any DC component off of the input signal. Its value isn’t very critical, but be sure to keep its value high if R1’s value stays the same in order to keep a low filter fc. This cap’s voltage tolerance must be higher than the maximum input voltage you expect, including the DC component. Since it’s being used as a filter, it needs to be as high a quality capacitor as you can find.
If your input signal has a high DC component, a fairly generic low-ESR electrolytic like Panasonic FC or Nichicon PW is probably the best you can manage. If your test signal’s DC component is 5 V or under, high-end large-value caps like Sanyo OSCON or AVX OxiCap are better.
Largest Part Size: 10mm diameter.
C2, C5, C9 (metallized polyester, ceramic)
These are bypass caps for the op-amps. 0.1 µF ceramic is traditional here; the same value in metallized polyester would also work.
Largest Part Size: 2.5mm × 10mm
C3, C6, C10 (tantalum)
These are reservoir caps for the op-amps. 1 µF tantalum is traditional here, but a 22 µF+ electrolytic may also be suitable.
If you are making the LNMP drive heavy loads, such as long cables, you might consider increasing the C10s to 10 µF. If you do this, you must use tantalums. There is no acceptable electrolytic replacement for such a large cap in this position on the board.
Largest Part Size: 4mm diameter
C4 (film)
This forms an RC low-pass filter along with R2 to limit the amplifier’s bandwidth. Some applications demand bandwidth limiting (e.g. power supply noise testing) and other times it’s simply a good idea to limit the bandwidth for stability reasons. The default value gives a 100 kHz -3 dB bandwidth point (fc).
If you add this cap for the purposes of setting a specific measurement bandwidth, I recommend the BC MKP416 series, available from Digi-Key. These are 2% metallized polypropylene capacitors, ideal for filter applications. They’re available in values down to 1000 pF, good for an fc of ~160 kHz or lower. If you need a higher fc or the exact fc isn’t all that critical to you, I recommend a metallized polyester box cap or an axial-lead C0G ceramic.
There are two ways to approach bandwidth limiting.
The standard way is to simply set C4 = 1/(2πfR), where f is the bandwidth limit and R is R2’s value. This puts the -3 dB point of the amplifier (fc) at f, the standard way of specifying bandwidth.
The other way is nonstandard, but it has the interesting property that it gives the same measurement over bandwidth that you’d get if you were able to implement a perfect brick-wall filter. It only works with signals that are fairly flat through the measurment bandwidth, such as typical linear regulator noise. The idea is, you set the fc such that the area under the RC filter curve is equal to the area you’d get under a brick-wall filter curve. For a simple 1-pole RC filter, the factor is 0.637 × fc, per section 7.21 in The Art of Electronics. For example, if you set C4 to 2400 pF, you get an fc of 64 kHz, and that has the same area under the curve as you’d get with a 100 kHz brick-wall filter. If you only have 2200 pF on hand, that’s pretty close, giving fc of about 71 kHz.
This part is optional. If you leave it out, the amplifier’s bandwidth will probably be somewhere around 5 MHz, all things considered, with the recommended op-amps.
Largest Part Size: 2.5mm × 10mm
C7, C8 (film)
These capacitors form a 2nd-order Butterworth high-pass filter along with R7, R8, and IC3. As they are filter parts, they should be film types.
The relatively small space set aside and the high default values were designed with the Wima MKS-2 series in mind. You can put up to 6.8 µF/50 V or 10 µF/16 V Wima MKS-2 caps in the space provided. If you go with another line, you will probably find that the largest value you can get here is about 4.7 µF, largely because most other metallized polyester box cap lines have higher voltage tolerances, which increases their bulk.
You should get enough of these caps to be able to hand-match a pair. The closer the two caps are matched to each other, the sharper the filter’s “knee.” The absolute value of the caps isn’t critical; they can probably be as low as 4.7 µF while keeping fc reasonably low. The only way to lower C7 and C8 further is to raise the values of R7 and R8, but that will raise the amp’s noise floor.
Largest Part Size: 7.5mm × 7.5mm
C11 (film or ceramic)
This is a compensation cap for IC3, to maintain good gain flatness into a capacitive load. (That being the cable going from the output of the amp to the measuremnt device.) Its type isn’t all that critical. Just use whatever you have on hand.
Largest Part Size: 2.5mm × 10mm
C12 (electrolytic)
This cap forms an RC high-pass filter along with R12. Its purpose is to strip any DC offsets that have accumulated through the amp. You’ll notice that none of the op-amps have offset trimming, and the input impedances aren’t even close to balanced; DC offsets will occur.
Its value isn’t all that critical. You just need to keep fc down in the single-digits range.
Since this is a filter capacitor, it should be a high-quality type. Since the voltage across it is low, you can use Sanyo OSCON or AVX OxiCaps here. At the very least, you should use a good low-ESR electrolytic, such as Panasonic FC or Nichicon PW.
Largest Part Size: 8mm diameter
C14, C15 (film or ceramic)
These are stabilizing caps for IC5. See the MC3334x datasheet for details. Cap quality isn’t critical here, so any old ceramic will work; polyester box caps are overkill on quality, but they’ll work, too.
Largest Part Size: 2.5mm × 10mm
C16 (electrolytic)
This is the amplifier’s main reservoir cap. Its value isn’t critical; something in the 100 µF to 1000 µF range will be fine. It should be a good low-ESR type if it is to do any good.
Largest Part Size: 10mm diameter
C17 (tantalum)
This is a noise-reduction cap for IC6. Its value is not critical, but it should have low ESR. A 1 µF tantalum works, but a 22 µF+ electrolytic may also be suitable.
Largest Part Size: 4mm diameter
Diodes
D1, D2, D3, D4
These diodes guide the currents in the battery charging circuit. Generic 1N400x types will work. 1N4001s are probably sufficient.
CLED
This is the charge state indicator LED. It will blink while the batteries are charging, it will be solid when the batteries are fully charged with wall power applied, and it will be off when the power switch is in the “amplifier run” position.
It needs to be a T-1 type if you’re going to solder it straight to the board. If you’re mounting it to the front panel, the type doesn’t matter.
See R18 and the MC3334x datasheet for further details.
PLED
This is the amplifier power indicator LED. See RLED for details.
The LED type isn’t critical. However, the board is designed for a T-1 type, to be soldered to the the board with enough lead length to bend the LED into position for a front panel mounting hole.
Integrated Circuits
IC1, IC2
These are the first and second stage op-amps. Because of the high gain of this amplifier and because these are the amp’s first stages, they should be low-noise types.
The preferred chip is the Analog Devices AD797, with only 0.9 nV/√Hz of noise. On paper the LT1028 and LT1115 should give similar performance, but they gave nearly the highest noise floor numbers in my tests for some reason; the only chips with more noise that I tested were the LM6171 and the OPA602. I found that the OPA228 gave the amp nearly as low a noise floor as the AD797; the 227 should give similar performance, but I didn’t test it because it only has 8 MHz bandwidth. My final choice would be the NE5534, whose noise floor was only 3 dB higher than that given by the AD797 and the OPA228.
There are no doubt other op-amps that will work fine here. I recommend a chip with at least 10 MHz to get a decent amount of feedback over a 100 kHz measurement bandwidth, and the specified equivalent input noise should be under 4 nV/√Hz.
IC3
This is the third stage op-amp. The standard part is the LT1206, a big cable-driver CFB op-amp. We use a cable driver since this amplifier will normally be used in conjunction with test equipment, and that means cables.
In addition to being the switchable 10×/1× gain stage, this amplifier implements a 2nd-order Butterworth high-pass filter. This isn’t strictly necessary, but the board space was there, so.... If you wanted to leave the filter out, you would jumper C7 and C8 and omit R7 and R8.
Linear Technology also offers the LT1210, a 35 MHz 1.1 A output current chip with the same pinout; it is similar in all other respects to the LT1206. This one isn’t carried by Digi-Key, so I see no good reason to use it unless your cables are many meters long.
The value of the passives surrounding the chip affect its performance more than for a VFB op-amp. You’ll want to study the LT1206 datasheet before adjusting any of these values.
IC4
This is an LM317 configured as a constant current source to force current into the batteries. You don’t need any of the premium variants of the 317 here, since it’s only used for charging the batteries.
There are LDO variants of the 317 (e.g. LM1086), but they’re tricky to make work in this circuit and there are plenty of downsides. The only reason to do it that your power supply’s voltage is just a bit too low to run an LM317, or because you’re striving for the absolute lowest heat dissipation.
IC5
This is the battery charge controller, On Semi’s MC3334x. Its job is to monitor the charge state of the battery, and to shut the regulator down when the battery is charged or when something goes wrong. It’s a complex little beast, so if you’re going to adjust anything about the charge circuit you’ll want to study the MC3334x datasheet.
Unfortunately, this chip is now obsolete, with scarce stock still available from the distributors. As I write this, the easiest source is the MC33342P from Mouser; this variant works best with slower charge rates.
Alternately, you can modify the charge subcircuit to be a simple but solid NiMH trickle charger like that in the PIMETA v2. Basically, you leave out all the parts in the charge controller subcircuit except for IC4, R14, D1, and D4, plus you jumper R15. The value of R14 controls the charge rate: R14 ≥ 1.25 ÷ I, where I should be one tenth the amp-hour rating of your battery or less. The recommended AAA NiMH cells are typically 800 mAh, so I = 0.08, giving a recommended R14 value of 16 Ω or more.
IC6
This is the Texas Instruments TLE2426CP “rail splitter”: it creates a virtual ground at ½ the battery’s voltage, allowing the op-amps to run as though from a dual voltage power supply. This is the low-noise DIP-8 version of the part, not the TO-92 version more commonly used in circuits on this site.
I considered tying the midpoint of the battery pack to circuit ground to avoid this part. The upside would be that I could eliminate this part and C17, and I could sink more current to ground. Since these advantages weren’t necessary, the downside was fatal: it adds another supply leg that has to be switched. You would need two switches (amp power on/off, charge on/off), and you’d have no way to avoid them both being on, which would have played havoc with the circuit. The only alternative that made any sense at all was to make the user remove the cells for charging; opening the case every few hours to charge the batteries is too big a pain to tolerate.
Enclosure
The LNMP is designed with the Hammond 1455L12 series case in mind. You can get these from Digi-Key, Mouser, and Allied Electronics.
If you’d rather get your parts elsewhere, the taller version of this case (1455N12) is more widely available. The extra height means the standard 8×AAA battery pack won’t fit snugly in it, but the charge circuit will let you charge a bigger pack just as well, either AAs or more AAAs. It does cost more, of course.
The Hammond 1455 series are rather expensive, so any old metal box will work if utility is the only consideration.
Don’t use a plastic case, because RFI getting into the circuit will wreck your measurements. With 60 dB gain, this is a very real possibility.
WARNING: The Hammond 1455 line has two panel options, plastic and aluminum. Twice over the course of the LNMP’s availability I’ve gotten reports from builders indicating that they chose the plastic option. While this is better than using a completely plastic case, I can’t come up with a justification for doing this on purpose.
Battery Holder
This circuit was designed to run from an 8-cell NiMH battery pack. This gives a nominal ~10 V supply, which sets the clipping point high enough for a ~1 Vrms output signal with a nearly depleted battery pack.
The board has a cutout for either one 8×AAA battery holder or two 4×AAA battery holders glued back-to-back. The cutout in the corner of the PCB is 2.1" × 2.1", and 0.1" on one side is taken by the rails in the side of the standard enclosure. That cutout and the height inside the standard enclosure are sufficient for the battery holders mentioned in the parts list. There’s a bit of rattle room in there, so you might want to get some thin foam to pad things out inside the case.
Power Supply
In order to charge an 8-cell NiMH battery pack, you need at least 16 V, assuming the regulator is an LM317. (This accounts for the voltage drops across D1, IC4, R14, D4 and the battery pack.) The power supply needs to have a current rating somewhat higher than the current you set R14 for, to account for efficiency losses.
The power supply can be either isolated or non-isolated. See the DC input jack discussion below for the considerations involved.
The unregulated supplies in the parts list are 12 V, but because they’re unregulated the voltage will actually be higher than this under less than a full load. The CUI Stack wall wart’s datasheet says it’s good for 16 V to about 0.3 A. The Xicon one is good for a bit more amperage, so it should manage 16 V at 0.3 A as well; it might even go higher, but since the datasheet doesn’t say, it’s a gamble.
Miscellaneous Hardware
The gain selector switch (S1) can either be an SPST toggle (or SPDT with one lug ignored) or a standard 2-pin 0.1" pitch jumper block. Use the toggle switch if you want to be able to change the gain from the front panel, and the jumper block if opening the amp case to change the gain is rare enough to be tolerable. You can leave the switch out entirely if you don’t want gain selection; just jumper it or leave it open as described in the Step-by-Step Assembly Guide.
The power switch (S2/S3) should be a DPDT type. You can wire this so that two of the lugs are connected in one switch position to run the battery charger from the wall supply, and two lugs are connected in the other switch position to run the amp from the batteries. (See the schematic for wiring instructions.) You could use two SPST switches instead, but your measurements’ integrity would probably be wrecked by the noise put out by the charger when both switches were on. By the same token, modifying the circuit so it only needs one SPST switch and the charger runs any time the supply is plugged in would also be bad. The DPDT switching scheme was developed to keep you out of trouble; use it.
The input and output jacks should be isolated BNC panel jacks. These are best for signal integrity, and they’re compatible with most test equipment. If you’re only going to use this amp with voltmeters, pairs of banana jacks would also work.
Assuming that the I/O connections are isolated from the case, the DC input jack should be a metal barrel type that connects the barrel of the plug to the chassis of the jack. With a tip-positive non-isolated supply, this connects the case to earth ground, which may be helpful for noise reasons. If your I/O jacks aren’t isolated from the case, you’ll need to isolate the DC input jack, due to the amp’s virtual ground scheme.
The two TO-220 parts (IC3 and IC4) have room around them for small heat sinks. You probably won’t need a heat sink on IC3; you’d have to be driving a near full-scale signals into a heavy load to need a heat sink. If the battery charging current and the wall voltage is low enough, you can get away without using a heat sink on IC4, but it’ll be a good idea in most cases.
The heat sink shouldn’t be much wider than the part itself and it can’t go down to board level. I recommend Aavid Thermalloy 577202B00000. There are common clip-on types that will also work. If you’re using the Hammond 1455L12 case, the heat sink can’t protrude much above the height of the part or it will touch the top of the case; the Aavid heat sink I recommended is a little too tall, but you can take off one set of fins with a hacksaw to fix this.
This space intentionally left blank. :)