Inuke 6000DSP filter mod guide for less low frequency rolloff
It has been a few months since i first did these filtermods for lower rolloff and the amp has run smoothly since.
I did say previously that i probably could cook up a guide for the filter mods if anyone is interested, just wanted to test it out some first.
Well, here it is, a guide and some random ramblings about the methods of finding these things out...
CAUTION, working with amplifier internals can be very dangerous, potentially leathal, so do be careful.
Do these mods at your own risk Obviously WARRANTY VOID doing these mods...
Since these amps drive my sealed subs I want them to have as little rolloff as possible, so I can get the most out of my subs when it comes to infrasonic output. I did some inspection on my inukes to find out where the low frequency cutoff / DC-block filters were and if they could be modified to provide less rolloff at lower frequencies. Obviously warranty will be voided doing mods like this, something to think about although I don't care, I consider the warranty voided anyway since I have done a fan-mod before... I just want better low frequency extention from my amps...
Since I did not find any schematics for this 6000DSP model, I had to use the schematics for 3000 without DSP as a general guideline. I then started working with a multimeter, probing for connections and drawing up some schematics of my own for the input signal path. I did not want to mess with anything on the "amplified" side of the board since I firstly don't really understand how it works and secondly it can be potentially leathal with its high voltage DC rails etc... So I decided to stay within the low level, non amplified part of the signal path, for safety reasons. It is also here that I suspected that any or atleast some DC-block / highpass filters would be.
After I had probed my way through a good portion of the signal path and drawn up a crude schematics I looked for tell-tale signs of highpass filters,
that is capacitors in series with the signal path. I found 3 areas of interest on the board, will go through each one at a time. I also did some rolloff measurements at these points to find out how much these filters affected the signals frequency response.
The rolloff measurements were taken with loopback measurements in REW,
feeding the Inuke amp with the subwoofer output of my HTPC soundcard that is direct coupled so it has virtually no rolloff to begin with. I should note that since the HTPC is my main source and I'm only interested in low frequency rolloff I let my usual 80Hz bass management filter be in place, so any rolloff measurements taken has first had the signal passed through this 80Hz lowpass filter. I used small measurement probe clamps with wires to attach to various points in the amps signal path and feeding that back to the soundcards Line-in input,
taking ground from a screw in the Inuke PCB. First though I checked if there were any DC voltage present with a multimeter at the meaurement points, if there were any DC present I took the loopback measurements with a large capacitor in line to remove the the DC voltage and protect the soundcard. I chose a 1000uF cap, thinking it would be large enough not to introduce any noticable rolloff on its own within the frequency span of interest. This were also confirmed with a loopback measurement of soundcard sub out to line in with the 1000uF cap in line, showing virtually no rolloff at all. Now, since REW can only display measurements down to 2Hz I thought that whould be a good place to take my measurements at, it is also about the lowest frequency that I guess could be of interest for the subwoofer range. In any case, if the loopback measurement show little to no rollof at 2Hz it has probably not rolled off that much even an octave lower at 1Hz... The way I took the loopback measurements was to make rew sweeps at fixed signal level and taking note of how much the signal level had dropped at the 2Hz lower limit in REW. With the measurement methods taken care of, lets move on to the actual mod at hand.
First an overview of the board in case someone don't know how an Inuke 6000DSP looks inside. I have also circled in red the 3 areas of interest.
The first filter I came across where right after the input signal buffer stage, after the XLR/TRS inputs. After the buffer stage the signal goes through a set of capacitors before it is sent to the volume atennuation pots at the front of the amplifier. These capacitors are on my board named "C20" for channel A and "C32" for channel B, and are a set of electrolytic SMD capacitors of the values 47uF and 50V. The caps and volume pots, when set fully clockwise (no atennuation) are responsible for 1.2dB rolloff at 2Hz,
not much but in my case I wanted as low rolloff as possible so I deemed the mod to be nessesary. Others may or may not want to do this based on your specific needs. If the volume knobs are not set fully clockwise the rollof should be a little more severe, I think, seeing as highpass filters work with a series capacitor and a shunt resistor to ground. The volume pot in this case should act as a variable resistor to ground, with less resistance the more signal atennuation you use. Less resistance to ground would mean that the highpass filter cutoff frequency moves up, in theory at least,
I did not test any other pot positions than fully clockwise as that is how I always use it.
I think these caps are in place to block DC voltage from getting to the volume pots as I've heard that DC present over volume pots could cause various noise when the pot is rotated. In my case I always use the pots fully clockwise (no atennuation) so in reality I don't need the pots at all, I control the input level to the amp upstream in the signal chain anyway.
I decided to remove the volume pots from the equation, I did this by pulling off the PCB connector labled "X4" that has the the wires to the pots running to the front panel. This exposed the bare connection pins on the PCB, as seen in the pic below. I also know that I do not have any DC offset present at the output of my upstream signal chain components, so no DC should enter the amp input anyway, unless it is encoded in the audio that is. I don't think low levels of DC here would be harmful but I don't know, however it will anyway be caught in another set of capacitors further down the line... Also I made sure to check for any DC offset present at any side of these caps, but there were none. Had there been DC offset present at either side of the caps they had to be inplace to block the DC from reaching the other side of the cap, and the components there.
With this in mind I decided to bypass the "C20" and "C32" caps alltogether and feeding the signal directly from the buffer stage to the pins on the "X4" PCB connector labled "A-VR-OUT" and "B-VR-OUT". That is the pins that the respective volume pots output to when the pot connector was in place. I did not want to do a permanent removal of the caps and volume pots, should I ever need to use the pots again,
instead I made it possible to change back to using the volume pots again. I did this by soldering wire pins to the small visible part of solderpad at the capacitors input side, then made wires to bridge these pins with each channels respective input pins labled "A-VR-OUT" and "B-VR-OUT" at the "X4" PCB connector. This way I can choose to bypass the capacitor and VR pot or removing these bridge wires, plugging the VR pot connector back on and using the volume pots like usual. I would do this mod only if you want to remove the pots, if you want to use them, keep it as stock, it doesn't give that much rolloff anyway...
See the picture below marking where to solder the wire pins and which pins on the connector to put the wires on.
I also include a picture of which types of wire pins and wires I used, they are sourced from a local electronics shop, found them under Arduino acessories... I just pulled each pin out of the strip and bent a small 90 degree angle at the bottom to give it more support for the solder. The wires came with female connectors at both ends, I just cut and soldered them to the nessesary lengths. I used the same type of pins and wires for the rest of the mods too.
Moving on to the next set of DC-blocking capacitor, these located at the signal input to the DSP module. The capacitors are electrolytic SMD ones of 47uF and 50V DC rating, 4 of them. They are on my board labled C133 and C134 for channel A, and C38, C39 for channel B. These capacitors cause a signal rolloff of about 4dB at 2Hz. They can not be bypassed since they have a DC offset of 2.5V at their output side, presumably the AD converter of the DSP has a 2.5V DC bias.
The way we can lessen the rollof are either to replace these capacitors with new ones of higher capacitance, or to wire new ones in parallell with the existing ones. If you wire them in parallell the resulting capacitance can be calculated through the formula CR=C1+C2 CR= resulting capacitance C1=capacitance of existing capacitor C2=capacitance of new capacitor
Whichever way you chose you should solder wirepins to each side of each capacitor. I chose to remove each capacitor to get access to the whole solderpad beneath the existing capacitors. As I do not have a hot air solder gun to desolder them with I decided to carefully wiggle each cap sideways, perpendicular to the solder pads until they came off. The idea is to weaken the material in the legs of the capacitors until they finally snap from fatigue. It is a fairly risky procedure where you do run a risk of ripping off the solderpads and traces from the board, so do be careful. If you don't want to risk that it is safer to solder wirepins to the small visible part of each solderpad instead like on the previous capacitors before the VR pots. The solderjoint can be fairly weak though seeing as it does not make contact across a large surface, so be careful when attaching the wires, so you don't break the solder joint. Perhaps cover the base of the pin in some hot glue or something to give it more support.
This is how the board looks after I had soldered the pins to the board. I'm not very good at soldering fine electronics though, might be among the worst actually
Afterwards you run wires from these pins to each leg of your capacitor of choice. Choose capacitors of the same or higher DC voltage rating (50V) and a capacitance you feel is right. I chose a capacitance of 470uF for all my replacement capacitors. I did loopback measurements and saw decreasing rolloff going from 100uF to 220uF and 470uF, with higher capacitance comes diminishing returns though. I also tested 1000uF caps but the gains did not amount to much, not worth the higher price and larger size for such large caps. Take note of the polarity of the original caps and wire up the caps with the same polarity as the original. The black markings on the original caps indicate the negative polarity side, that is the input side of the caps. I then duct-taped the capacitors to the top of the DSP box, so the caps stay in place and relieve the wirepins of any stress.
Then there is the last DC-blocking capacitors, located after the DSP, at the input to the amplifier stage of the board. The capacitors are ceramic SMD ones labled on the board as "C45" for channel A and "C46" for channel B. These have no markings so I do not know any values on these ones, they do however look about the same size as the ones on my 3000DSP amp. I do not know the values on those either but I do have schematics for 3000 non DSP, there they are labled as 10uF. Assuming Beringer used the same ones on 3000 and 3000DSP they could possibly be 10uF on the 6000DSP too.... By my measurements these caps are responsible for about 14.5dB rolloff at 2Hz.
I did measure about 35mV DC offset at the input side of these caps, no DC offset after so I thought it would be wise to keep the caps and not bypass them. I desoldered them, soldered wirepins on either solderpad and wiring up replacement capacitors instead, also 470uF 50VDC electrolytic ones were chosen here. I saw diminishing returns of 1000uF caps so 470uF it was, easy to buy the same caps for all the filtermods. If desoldering the ceramic caps seems scary one could solder wirepins directly to each end of the original ones, and wiring up the new cap in parallell with the old one. As always care has to be taken not to damage the solderpads or traces on the board... If it matters I chose the same polarity wiring as for the DSP filter mod, that is input signal on the negative polarity leg of the cap and output on the positive.
A pic of the soldered wirepins for these caps, pardon my poor soldering skills again, also a pic of the complete job, with caps wired up. Not as tidy as the stock board layout but it gets the job done...
Thats it for the mod, after these 3 filtermods the signal after the "C45"/"C46" caps measured as the blue trace below, 470uF caps all around. It shows little to no rolloff at all at 2Hz, do not know where the "squiggly" signal comes from, it came and went on a couple of measurements, probably measurement error or something with the line-in calibration file. The red trace are the signal after the stock DSP input coupling cap, taken through a 1000uF cap to filter out the 2.5V DC offset. It shows the signal rolloff from the stock caps at the DSP input. The green trace shows the signal after the "C45"/"C46" with stock caps at ampstage input and DSP input, but with the input coupling cap before the VR pot already bypassed. That mod I did before anything else, still that was only about 1.2dB rolloff at 2Hz so it does not have a large inpact on the figures...
I also took measurements from the signal input pins of the PWM class-d chip, through a 1000uF cap to filter out DC offset and protect the sound card line input. The signal rolloff matched the blue trace at the above pic. That is not much rolloff at all at 2Hz. I cannot take measurement from the amplifier post signal amplification but I did some visual inspections of the subwoofer cone excursion when fed a signal from a signal generator. My findings were that the cone excursion did not begin to fall off until about 2Hz, and some but not large reduction in excursion at 1Hz. That strengthen my belief that these filters I have modded are the only highpass filters prescent on these amps, filters worth modifying that is. I see little point of the amp running flat out to DC, a Hertz or so from DC is where i jump off the train when it comes to neccesary extension for subwoofer amps... That is it, a guide to mod a Inuke 6000DSP to have "about flat" frequency response down to 2Hz. Might be something to consider if you are so inclined, are modding friendly and don't care about warranty, just want a "flat" amp to power some sealed subs "all the way down"... Of course I cannot guarantee that these mods are in any way safe to perform for the amp or other equipments for that matter. That being said I have run my amp modded like this for 4 months now and I haven't noticed anything bad or any odd behaviors signaling the amp might not like it. Only thing I noticed was A LOT more low end infrasonics coming from the subs when it is prescent in the audio