DM500 audio mod

An excellent tut frm gw1 @ austech

A friend of mine who owns several receivers including a DM500 told me he's sick of getting blasted by loud volume when he switches from DM500 to his other equipment. Besides setting up an external amplifier for his Dream, what else can he do?

Approach
This issue gets raised regularly. Z80 kindly contributed DM500 clone schematics and suggested modifying the filter to add gain. The problem with that approach, apart from the considerable difficulty of cutting tracks or lifting leads without causing damage, is you'd need to adjust a bunch of resistors and capacitors. ie you can't change a filter's gain without recalculating the RC networks otherwise the rolloff frequency shifts and you have phase distortion and potential instability or clipping to contend with.

The Dream engineers needed a solution which was cheap and compact. DM500's audio DAC (PCM1725 or WM8761 depending on board revision) has output amplitude proportional to supply voltage. Dream use 3.3V, unlike many older receivers which use 5V, but kept the simple unity gain Sallen-Key filter configuration used by many other receivers rather than employ a larger and more costly multistage one with gain. In short that's why the DM500 is quieter than other receivers.

Just how much quieter? As you can see the DM500s have only one-third the RMS amplitude of Humax, Strong, UEC and Manhattan. That's not the same as one third the loudness, but it's still a very noticeable difference.



The above are AC RMS measurements I made using a Fluke multimeter, Optus C1 TUNE channel's 1kHz -14dbVfs sine audio, 47K load on receiver output, set to maximum volume. It's not an ideal test setup but it answers my question: it shows I need about 0.31/0.122=2.54 times extra gain.

There are several ways to get it. I wanted something that was internal (no extra power supply, cables or dongles hanging off the back), made from parts stocked by local suppliers, easy to build and install, and with reasonable performance - no oscillation or noise problems. I prototyped a filter rework option but in the end decided to keep the existing one and simply add an amplifier at the output. The main advantage of this approach is simple installation: you just need to desolder two resistors and tack on a few wires in easy-to-access positions. Anyone with reasonable soldering skills and a 1mm iron tip should be able to do this without trouble.

Circuit Description
The circuit is shown below. Looking through local supplier catalogs I chose the LM833N op-amp as it's a dual one with decent performance and lower noise than some of the really cheap ones. Audiophiles wouldn't use it but it's on par with most budget audio gear. I've set the gain at 5K6/2K2=2.54.



The full-scale DAC output is 0.62*3.3=2.05Vpp, which means our amplifier output needs 2.05*2.54=5.2V peak to peak swing. As LM833 isn't a rail-to-rail type (they're harder to find locally) it needs 1.5-2V clearance at each rail to avoid distortion. That means we'll need a clean 5.2+2*2V=9V supply. The DM500 has no clean high rails with adequate current so that means an extra regulator is needed. I chose a 7809 rather than LM317 to minimise parts; its TO-220 package is bulkier but space isn't a problem, it's readily available and cheap. By using the inverting amplifier configuration I avoid having to buffer the midrail reference.

As the LM833's 4.5V DC operating point is above the 1.65V of the filter, C116/C117 couldn't be used due to their polarity, hence additional DC blocking capacitors on the amplifier input were necessary. C116/C117 couldn't be reused for output DC blocking either without cutting tracks, so I chose to intercept at R110/R111 and add current limiting resistors at the amplifier.

The schematic doesn't show the RCA audio output jacks, but they're directly in parallel with their respective SCART pins. So the increased volume will be available at the RCA jacks too - you don't need to be using SCART. I focus on SCART only because its connector is the easiest place to attach wires.

Power can be obtained from DM500 12V reservoir capacitor C422. Normally you'd take analog ground from the DAC or filter, not somewhere further back, but in this case the two are quite close on the DM500 ground plane so there's no earth loop problem.

There's no absolute standard output voltage for line level audio, just conventions for different industry segments. You can tweak the amplifier gain if you wish. For slightly higher gain increase the 5K6 to 6K2, but don't go too high or you'll approach the point of clipping. Don't reduce the 2K2 too much or you'll overload the filter and reduce its performance; don't increase the 2K2 above 10K either or you'll needlessly amplify input current noise. I think the given values are about right.

Parts List

Construction
The circuit can be constructed easily on veroboard. Cut it to size and make the 10 track cuts using a drill bit before mounting any components (inspect carefully with a magnifying glass to ensure fully cut). Fit the two ground wire links next, then the IC, then the remaining wire link between IC pins 3 & 5 on the copper side of the board. Then mount the regulator, the resistors, the ceramics, and lastly the electrolytics (take care to observe correct polarity). Note that one lead of the Lout capacitor passes through a drilled hole; solder it to the adjacent track as shown. When finished make a cuppa and spend ten minutes checking component positioning and inspecting solder joints with a magnifying glass as there's a fair chance you've made at least one mistake.



The veroboard can be secured to the smartcard socket with double-sided tape but don't do that until it's fully tested and working.

To remove the two resistors from your DM500 you will need a good lamp and magnifying glass (an inexpensive 10x jeweler's loup such as those on [ame="http://www.amazon.com/Jewelers-Loup-Times-Magnification-Loupes/dp/B000E4Z3Z4"]ebay[/ame] is ideal), a fine soldering iron tip (0.5-1mm), fine solder (0.5-0.8mm) and a damp sponge for tip cleaning. Ideally you'll have some [ame="http://www.altronics.com.au/index.asp?area=item&id=H1650"]flux[/ame] too but you can get by with solder provided it's high quality low-melting-point type and your iron isn't too hot.

The resistors to be removed are R110 and R111, indicated below. You don't need to remove the motherboard but you may find it easier if you do.



Important: If the component layout on your motherboard looks different then your circuit may be different. You should trace from SCART pins 1 & 2 back though the current limiting resistors and electrolytic blocking capacitors to the amplifier output pins. Don't desolder anything until you know exactly where to tap the filter output and which resistors to remove.

To loosen each of the resistors put flux over each one if you have it and heat each end. If you don't have flux then wipe iron tip clean, heat each end briefly and apply the tiniest amount of solder to each joint to lubricate it. Then after momentarily heating each end you should be able to dislodge and remove the resistor very gently with your tip. While removing it be careful to keep it away from other components to avoid a short circuit. Don't apply too much pressure with your tip or you risk lifting tracks off the board. Apply only a tiny amount of solder or you'll end up with a mess. Don't use desoldering guns or solder wick on the DM500 board or you'll risk track damage. To remove excess solder wipe your tip clean on a wet sponge then mop up gently with tip, being careful not to apply excessive heat.

With the resistors removed, inspect the board carefully under a good light to check you haven't left any blobs or splashes of solder anywhere which could cause a short circuit.

Prepare the six wires, cutting to the required length, stripping, lightly tinning and soldering onto the veroboard only at this stage. The leads for Lin and Rin are the most critical: use only thin flexible wire (I used IDC wire), twisting tightly to avoid whiskers.



Installation and Testing
If you chose to remove your motherboard from its case you should clear your work area before applying power: ensure there aren't any metal objects or offcut component leads on the bench which could create shorts.

Always unplug all cables from your dreambox before touching it with your soldering iron! There's a risk of damage if you fail to do this. Also you should take the usual static electricity precautions.

Connect the GND and +12V wires only to begin with. I suggest either side of C422 since they're easy to attach to. Don't apply excessive heat though: you don't want to damage C422 or its PCB pads. Double-check you have polarity the right way around, then apply power to your receiver. With your multimeter's black lead on GND (eg the 7809 tab), check you have about 9V on IC pin 8, and 4.5V on pin 5 and pin 3. If not unplug power immediately and check the connections around the regulator and the 10K resistors.

Next check that you have 4.5V on pin 1 and pin 7. If not then remove power and check the connections around the 2K2 and 5K6 resistors. Look for short circuits or improperly cut tracks.

If the voltages are OK remove power and solder the remaining four wires to the motherboard. Prior to attaching wires to C131 and C134 you should prepare the connection points by heating gently and applying a tiny amount of solder. Do not use excessive heat or you risk damaging the capacitors. Avoid too much solder as it may create a short circuit. Check your work thoroughly before reapplying power.

With a multimeter you should find that you still have 4.5V on pin 1 and pin 7. Tune to a suitable program, crank up the volume to maximum and check that the sound isn't distorting. For testing I use a spare 47K resistor soldered across an RCA plug, insert into one channel at a time and measure AC RMS voltage while tuned to Optus C1 TUNE at maximum volume. If you have a CRO you should observe a sine wave about 0.31Vrms*2.83=0.87V peak-to-peak. Other programs will have considerably higher peak amplitudes, but you should never observe clipping. If you do experience clipping or distortion that would suggest incorrect component values or bad solder joints.

Here's a shot of the genuine DM500 2004 V6.0. As you can see it has identical layout to the clone I described. It uses the Wolfram DAC which, unlike the PCM1725 used in some clones, is designed to work properly at 3V.

This clone is similar but inferior for several reasons.

• Its PT8211 DAC component is much cheaper because it contains no digital filter. Consequently because the receiver has only a third order filter the audio quality won't be as good, it'll have more high frequency noise.
• Its LM358 op-amp is very basic, one of the cheapest you can get, with ordinary noise and gain bandwidth product figures. Unlike TLC2272 used elsewhere the LM358 isn't a rail-to-rail type which means it may have noticeable distortion on louder programs.

I haven't tested on this receiver (I only have an old carcass) but suggested attachment points are shown below. Parts are missing because I was pulling it to bits, not because of the mod!



Alternative desoldering option for V1.2 clone
Resistors R110/R111 are harder to access on this receiver, nestled between the SCART connector and electrolytics C116/C117. The resistors are in series with those capacitors, so you have the option of removing the capacitors instead. To desolder C116 and C117 you'll need to remove the motherboard from the case. Turn it over, then for each capacitor heat both of its leads, applying extra solder to carry the heat through the hole. With both joints molten gently wiggle the capacitor out. Don't pull with force though or you'll rip the plating out of the holes and/or cause track damage. They should come out easily once sufficient heat is applied on both leads.

So, to be clear, if you desolder C116 and C117 you don't need to desolder R110 and R111. The wire connection points remain the same regardless of which desoldering option you choose.

Reduce gain for V1.2 clone
It turns out the PT8211 has different output level to PCM1725 and WM8761 which means you may not need as much gain when modding this receiver. The DACs described earlier at 3.3V have full-scale output swing of 0.62*3.3V=2.05V peak to peak. The PT8211 however at 5V has typical swing of between 2.2 and 2.7V peak to peak, 2.5V being typical, according to its data sheet. That's a hellava wide variation, mainly with temperature I suppose, which is why it's cheap. But it's more than the 2.05V output of the others so less gain is required.

Here's how to work out the new resistor values. Based on RMS output observations I decided to boost gain of PCM1725 output by 2.54. That means our target full scale amplitude is 2.05*2.54=5.2V peak to peak. The PT8211 will therefore need gain 5.2V/2.7V=1.92 minimum, 5.2V/2.5V=2.08 typical or 5.2V/2.2V=2.36 maximum. That means changing the 5K6 resistors to either 4K3 (1.92*2200R), 4K7 or 5K1. I suggest trying 4K7 first and see what you think. If you have Optus C1 and an AC multimeter you could measure on the audio output jack when TUNE channel is cranked to maximum volume; a reading of 0.30-0.35 means you're in business.

On the other hand, if you don't change the resistors but keep the 5K6 then the amplifier output will typically swing 2.5*2.54=6.4Vp-p, or perhaps up to 2.7*2.54=6.9V. That's right on the margin of headroom needed with a 9V rail, so chances are the LM833 wouldn't be clipping but it might on full volume. I'd be more concerned though with the RMS output which may exceed the specification of your other equipment (eg DVD recorder or hi fi amplifier) and cause distortion there.

So why cant you just lift pin 8 VDD and supply the chip with 5V??

I have a 2004 rev 7.0 and it uses a PT8211 for the DAC and that is supplied with 5v.


The digital audio signals from the ARM SOC (serial data, bit clock, LR sample clock, and system clock for the digital filter) are all 3.3V level. Most CMOS devices need their inputs to swing properly between rails. If you power your DAC at 5V you need to give it full-swing 5V inputs. It may have TTL or CMOS switching characteristics but it'll still want full swing.

The PCM1725 is one such device. It is 5V CMOS with TTL compatible inputs, designed for 5V logic input. It may tolerate 3V level input but it's quite possible it may draw excessive current or misbehave when used that way. That's why even the cloners who chose to use that part didn't power it at 5V. They gave it a 3.3V supply, even though the device doesn't officially support that lower voltage, because that's safer than driving its digital inputs incorrectly.

Your 2004 V7 clone uses the PT8211 because it's cheap, and they're able to do so safely because it is specially designed for 3V bus compatibility when run at 5V. Its datasheet highlights that as a major feature. The genuine Dreams don't use the PT8211 though, even though it would give louder audio output, because the other DACs have interpolation filtering and noise shaping which gives vastly superior SNR and THD performance.

In general when you need to drive 5V logic with a 3V signal you need a level shifter of some kind. If the driver is rugged you can sometimes get away with using pullups, but you definitely don't want to do that with the ARM SOC. Sometimes you can use open collector/open drain buffers, though that's not the best solution for higher frequencies (and the PCM1725 system clock is over 12MHz). There are ways to make the DAC work with 3V bus if you really needed to, but not simply with locally available parts and without cutting any tracks. Accepting the low DAC output and either changing the filter or adding an amplifier is easier and much safer.

My documentation above has incorrect SCART pin numbering. My build and photos were correct, it's just the writeup that's incorrect

• On the circuit diagram, the SCART pin marked "2" should read "3"
• On the veroboard component overlay, the wire marked "Lout SCART 2" should read "Lout SCART 3"
• The comment after the connection point diagram should read "Important: If the component layout on your motherboard looks different then your circuit may be different. You should trace from SCART pins 1 & 3 back through the current limiting resistors..."

I can't correct the earlier posts due to forum edit policy. However a more compact version for DM600 PVR is coming which will show the corrections.

Here is an alternative layout. Same circuit, same connection points on the motherboard, just slightly more compact.

cont from last post