SSL-2 audio interface

[Tubetec]

I made up a battery supply , 3xAA alkalines in series ,
the weakness of the current supplying capacity of the battery showed itself as even earlier breakdown into oscillation on hi-z mode , the dominant discordant note in the intermodulated mess that resulted was gone .

On mic input even better results , not really a measurable difference in overall averaged noise level , but a more white noise like backround mush , more random , less fizzly with harmonics .

I dug out a 57 for comparative purposes , up close through a flat eq'd channel as you'd expect , a little woolly in the bass , still crisp and clean in the top end though and extremely low noise .

There is a couple of headers on the back edge of the pcb , one marked JTag one PWR ,
I'll have a poke around with the meter tomorrow and note down what I find.

Would be interesting to find out from anyone else if the SSL2 willingly goes into oscillation on high -z mode with the gain cranked and nothing connected , maybe its a foible with my particular unit ,
In any case audio equipment with nothing connected should never break into spurious osscillation at any setting of the controls , typically in a guitar amp input is grounded as soon as the jack dissconnects , you would think a Neutrik combo socket does the same by default ,

Theres a few weird things I discovered , if I make contact with the underside of the hi-z switch or terminals on the input socket the oscillation momentairliy goes away , having Mhz bandwidtch on a DI input is questionable .
I might try a few pF's to ground across the unbalanced input jack to try and neutralise the misbehaviour ,

 

[Tubetec]

I found a report detailing the same issue on an SSL2+


'It appears that the SSL 2+ mainly results in noise when having the "Line" enabled on the guitar input (with the gain dialed to "11" on input 2).

The SSL 2+ has far less noise present on Input 1: (with gain dialed to "11" on input 1).

Again like mine its like channel 2 is the source of the trouble .'

The results with the 3 AA batteries were good from the point of view of noise , but AA cells are a bit small to deal with the current demand , so the voltage on load drops to around 4V from 4.8v ,

I was thinking of trying four nicad cells for 4.8V
can I supply +2.4v and -2.4v to the device on the red and black usb wires and connect the 0v(cable sheild) to the mid point in the battery ? would the interface even power up , or would it damage something ?

 

[Tubetec]

Ive tried various means to supply the interface power , anywhere below around 4v or above 6v and the unit wont power up .
Noise floor is a little cleaner running on batteries , the lack of current from a small battery reduces headroom and also causes further instabillity with high gains on the hi-z input
I measured the point in the circuit where its marked as a 15 volt rail , I only see around 8 volts , that seems to explain the +10dbu max output ,
Id imagine the SSL12 interface is substantially the same with the addition of two extra inputs , but with a better power supply it has an ouptut capable of +18.5dbu .

 

[Tubetec]

There was a question someway back in this thread about low input impedence , ie 1200 ohms at the mic input .
I found the noise performance with an SM57 to be outstandingly quiet , best Ive ever heard .
If I was feeding a tube mic into this interface Id almost certainly use the line in with 20k ohms ,
you still have all the gain you need on tap with less load imposed on the mic ,so it will handle the peaks with less (obvious)distortion .

 

[Tubetec]

Oh well , I somehow managed to blow the switching IC for the USB power , its the device that senses the input voltage and decides to start the device up or not , its probably the same thing that causes USB bus powered interfaces to flicker on power up if theres just less than enough juice to start it, in the case of overvoltages it cuts out the power .

Anyway its over and done with now , I was able to bridge two test points on the pcb marked '5V bus' and '5V SW' to bring it back to life , damn that was a close one !
I dont wanna ham this thing up , I best be extra carefull about what voltage I feed it from now on .

I tried 3 cells of 6V lamp battery , giving around 5V unloaded , around 4.5v on load , its a little bit later when the screech kicks in , but very much the same , woefully inadequate in terms of current capabillity ,

Just haveing a good nose around the circuit now , the incomming 5v rail is split , one line to the switching IC which I blew up , that in turn feeds the analog circuits , another line goes to 3.3V and 1V regulators , it looks easy now that I blew the switch, to feed both analog and digital rails from a different 5v source , maybe its the interaction between the two thats causing the noise which is definately digital in nature .

I could actualy live with this box being a simple pair of mic pres into a mix bus and give up the digtal functionality entirely if it takes its digital garbage away with it .

Next I'll try powering up only the analog section and see if the digital noise goes away or not .

 

[Tubetec]

I stopped concentrating on the supply and decided to add a layer of conductive aluminium tape on the inside of the plastic tray that houses the pcb ,
I made sure it makes good contact all along the edge of the top plate and back panel ,it also has a connection to the base plate , it provides continuous screening around all the electronics .
Its made the best improvement yet and Im still only running off a single rail supply ,
Theres a marked reduction in HF garbage comming through now , the point on the gain control the instrument input breaks away into oscillation is only in the last 5% of the dial towards the maxiumum position ,thats a big improvement from what it was .

Theres other unwanted interactions though , still some residual whine although much reduced from what it was , I might have another go at seperate analog and digital 5V rails tomorrow as it seemed to bring about its own improvement when I tried it before the screening job .

The dry cell battery I got has run down to the point I was able to run the four cells to get a healthy 5.2v on load ,which made an incremental improvement over the previous 3 cells which in got me around 4.5V on load . Laying off the digital supply to the USB connector should extend the life of the cells another bit ,
I have a bag of nasty Nicads cells 20 years old , I might be able to select out a few for revival .
They would provide the Analog stages with huge power reserves , thats bound to improve the performance a bit especially when the headphone amp is driven .

As it was a fine day today I did most of the testing on the bench out back , I decided to provide a local earth appart from anything to do with the utillity company , Im running off batteries in any case so its not a matter of safety . I made the temporary ground with a large concrete fixing bolt , corkscrewed down into the soil . Then I used heavy gauge copper and referenced the various places to earth , chassis -negative battery terminal-ground gave a good reduction in noise , referencing postive battery terminal to earth induced some extra noise ,
The USB cable screen was also creating a loop back to chassis , I'll have another look tomorrow and see if I can refine things a bit more ,

 

[Tubetec]

Now that Ive reduced the dominant sources of noise a good bit , it reveals other sources , its impossible to test the noise with the unit open , theres simply too much garbage picked up ,
Probably time I add a dedicated power connector to the back panel , like any grown up interface has ,
impossible to test with crocadile clips and wires down onto the board .

I tried again today , single rail via USB battery bank , I grounded the spiral metal sheilded usb cable at the powerbank output , it reduced the tendency to oscillate by another few db on the scale ,
The powerbank is only in its first few charge cycles so I suspect some of what Im hearing is due to it setting down , there is also a variance in the amount of noise induced as the battery runs lower and the regulation circuit has less voltage to play with .
Thats not something you want .

Nicad power is the next step .

 

[ccaudle]

Nicad power is the next step .

Did you consider 6V SLA? At least in North America that voltage is used a lot for power backup for house alarm systems, and some small children's ride-on toys (small electric cars). 6V is a little bit low to regulate down to 5V, so it might be in that in-between range where it is too high to use directly, and too low to regulate effectively.

 

[Tubetec]

I persisted a bit longer experimenting with various 0v and grounding arrangements .
It turned out the metal sheild on the USB cable wasnt referenced to the plug , I made a metal loop for the usb cable to pass through ,bonded to chassis , I also used the conductive foil to make contact with plug shell and sheild , both those two things helped a lot ,but the break through came when the outer shield of the usb cable touched the metal headphone jack , all of a sudden another large reduction in steady state noise and the whirely squirely oscillation only happening at the absolute max gain position now.

I doubled back then and started adding a ground connection in various places , applying ground anywhere now no long has any influence on the noise level .
Were stil not at pure white noise yet .
The ad hock arrangement where the usb sheild touches the audio 0V on the headphone jack isnt optimal , it creates its own small loop , tomorrow I'll try and tidy it up a bit , minimise the loop and hopefully the pig sqeal will be gone as well .

Theres something fishy about the way the USB and audio ground is done on this board , but Im glad I went ahead and did the extra screening of the tray in anycase ,

When I started out anything above 7 on the dial with an open Hi-Z input and the thing was screaching at me in the cans at any level , now its only at the very last increment on the gain pot ,with headphone out at full before chaos breaks out , gettin' there ....shortening the inductive loop,

 

[Tubetec]

Theres no doubt an old fashioned 6V SLA or GEL lead acid would do it but I still seem to be making improvements with just the USB powerbank for the moment ,
In the end of the day I think it comes down to the internal impedence of your supply determines how well it silences
The dry cell battery in itself doesnt produce much noise , but its rising internal impedence as it discharges allows noise from the SSL2 digital supp;y to appear in the audio path , the usb powerbank is the same as the charge level depletes , the harmonic spikes come up in the noise floor .

I havent even had the chance to hook this unit to a pc yet appart from bus power , its windows 10 or above ,its just me ,the DMM, a set of cans and a 0v probe wire ,it doesnt need to be anymore complicated than that .

 

[Tubetec]

It seems likely to me the difference in performance between the SSL 2 and the SSL12 is largely due to the powersupply implementation ,

 

[HP200CD]

I injected a 440hz balanced test signal from the iphone

How did you get a balanced signal out of an iPhone?

its windows 10 or above

My SSL2 works fine on Windows 7

 

[Tubetec]

Hi HP ,
So you have the SSL2 as well , have you noticed the instabillity on the hi-Z inputs ?
I tried to first install the windows driver from SSL ,It came back with a message saying the installation wasnt meant for my OS , I then tried the firmware update package ,which came back with the same error .


What I did was used Reaper to make up special test tones , in wav file format .
You flip the phase of the right channel test tone relavtive to the left side , otherwise the signal is identical , now you a get balanced output connected between the two headphone outputs ,
I made up a cable using an old set of Apple 3.5mm ear buds with the remote , you simply remove the drivers and replace them with a male XLR plug, Lch to pin 2 ,Rch to pin 3 , now you have a balanced signal out , the remote also works for play/pause in the media player and volume control in nice repeatable steps .Its my new best friend this gadget , It took time to dissable all Apples add ons , but now it just does what I tell it , no phone calls ,no reminders ,

With this technique its possible to generate a single balanced signal from any standard two channel phono input or output . By the same token its possible to convert a stereo amp into mono by feeding L/R channels with opposite phases , then taking your output between the L and R positive terminals .

It seems to me the 0V connection on the SSL2 headphone output isnt referenced to anything , as you read above I found out inadvertantly by shorting the metal headphone jack barrel to the chassis , it cuts down on a lot of the induced noise and interactions between Hi-z input and headphone out.

I just had a check on the lead acid 6V batteries , it would cost me at least 50 euros for a Yuasa motorbike battery , I can get 4.8V/4000mah of Nicad emergency lighting battery for 12 euros ,
https://batteryworld.ie/products/nicd-4-8v-emergency-lighting-d
Thats a fair bit more convienient than lead acid , plus the voltage is more suited , Nicads as we know were commonly used in cordless tools and drills and can take serious punishment , the 500ma or so the SSL2 needs to run will make life very easy for the cells and inconjunction with a microcontrolled charging routine should give very stable long life . Another benefit of Nicad is they dont sag ,they deliver most of their output power over a narrow voltage range ,then the voltage falls off a cliff at the end . Its easy to incorporate a led voltage display to keep an eye on things .

 

[Tubetec]

I got two of the nicad emergency lighting batteries , 4.8v4000mah.
they started off around 3.75V off the shelf , on charge they went upto around 6V , off charge off load they dropped back to around 5.5V I expect them to settle at around 5.2V on load .

Im a little worried since I killed the regulator at the input about applying nicad power ,
It turns out its a linear regulator , at 6V input it cuts out , otherwise delivers around 4.5v regulated at the output to the analog buck boost stages .

I remember back to when I used run 1/10th scale nicad powered RC cars ,
they used a 7.2V 1200mah pack , it got fast charged in around 20 minutes , you could get around 5-10 minutes running time . The electronic speed control could deliver upto 25A .

Now I have over 3 times that capacity and very much lower internal resistance in the D cells , Im worried if I connect it the interface will go up in a puff of smoke .

I did a bit of checking around and it seems a few others have encountered the blown regulator also , the literature states 6V maximum , I might have exceeded that with the dry cell .

Any precautions that are worth taking before I apply the power ?
I was thinking of applying an external load to the batteries first to reduce the off charge voltage closer to 5V before applying the power . Any advice on this very welcome .

I dont want to turn my nice new SSL interface into a doorstop , Ive seen first hand the power of nicads when a direct short across the power connector developed in my RC car , luckily I was able to get the fused battery out in time or I would have been left with nothing only a burning pile of plastic . the tabs connecting the cells turned white hot in a few seconds while hundreds of amps of current flowed , beware the power .

In reality Nicads are almost the perfect power source , ultra low internal resistance which means AC noise voltages are very effectively supressed , If I wanted to make a mains PSU that could deliver 5V @25A I'd probably need a transformer of a couple of hundred VA , not to mention what would happen to the voltage off load .

The standard thin wires in a USB cable I also expect allow noise voltages develop , a decent power connector might also make a difference in that respect ,USB power in so many ways is suboptimal for audio.
Its not so much the fact that its only 5V ,its the grounding becomes a nest of two headed hissing snakes .

 

[HP200CD]

So you have the SSL2 as well , have you noticed the instabillity on the hi-Z inputs ?

I don't use the Hi-Z input setting because the SSL2 is not part of my recording studio. I use the SSL2 in my home office for Zoom calls. Mine is powered from the USB cable. I did just now test both inputs on the Hi-Z setting and could not create any oscillation, even at maximum gain with the input open circuit. What I did notice is digital noise, probably coming up the USB cable from the computer. So yes, the Hi-Z input is not hi quality. I would ignore it and use a good DI box on the line input. Even at maximum gain, the noise on the line input is under -100 dbfs and at minimum gain is under -120 dbfs. Only a few years ago those noise figures were only possible on the most expensive A/D converters.

Additionally, let me say that open circuit noise measurements are pointless, because nobody records an open circuit. Noise measurements are typically made using a load resistor of the appropriate value for the circuit under test across the input.

Please look at all the attached open circuit noise spectra (Hi-z, line, and mic, each at both minimum and maximum gain)
These charts were made with REW (Room Eq Wizard) which is free software.

Considering the low noise on the mic and line inputs, I think the SSL2 is a very good value, especially for $200 new. Quit nitpicking and make music. Enjoy.

 

[HP200CD]

everything is great until you hit a little above the 0db segment on the meters then very abruptly over the space of a few db it generates a range of audible harmonics , its nasty stuff , defiantely a place you dont want to be going levelwise .

All A/D converters will clip at 0 dbfs by definition, unless some analog limiting is built in to prevent reaching 0 dbfs.

Generally live music is recorded at -20 dbfs or lower (avg or rms level) so that the peaks remain under full scale.

 

[Tubetec]

Thanks for taking the time to do the tests HP,
I'm quite well aware an open circuit noise test on hi-z doesnt represent a real world situation . I did of course test it with the input shorted and with resistors of various values .also .
Did you try the test with the headphone volume up full ? thats where the instabillity kicks in worst in mine .Again its not a likely scenario in the real world , but Im looking at worst case situations ,where for instance the 2nd channel is on hi-z with the gain up and nothing connected and contributing noise spikes to the mix bus ,usually on hi-z inputs the jack is shorting , this isnt the case with the SSL2 , perhaps its due to the combo socket , I notice on the SSL12 the hi-z inputs are available on shorting jacks at the front , the neutrik combo is dedicated to mic/line.

Shorting the headphone output 0V to the back panel , you'll see those noise spikes are very much reduced , theres still a trace of the noise left, the actual amount depends to some degree on the quailty of the 5V you give the interface

I was talking about 0db on the meters and distortion only in the analog path , as I said I havent been able to get my unit working with Win 7 yet ,
If you've managed to get it working with Win7 you might let me know which version of the driver you used , because the one I tried wasnt compatible ,v5.30.0.13
I sometimes use a program called Driver Double to extract drivers from an OS , maybe you could pull the SSL driver from your Win 7 and I'll give it a try on my system .

The mic and line inputs do give suberb noise performance but those spikes are still just visible in the noise floor .
I know Im being pernickety but I think theres room for improvement of the grounding arrangements and power supply .

 

[Tubetec]

Its the driver version 4.67 that works with Win7 ,
they dont make that clear on the website ,

The firmware updater seems only to be compatible with Win 10 or above .

Single rail battery power still has the same gremlins unless the outputs (headphone and line) have their 0V connected to chassis , then the motorboating vanishes , but theres still some trace of instabillity towards the top end of the gain control , it sounds like poor decoupling between stages .

Tomorrow I'll try dual 5V rails ,seperating digital and analog lines and see if that has any impact on the howlround . These units are only out three years , the one I have isnt that old , be a shame if these things go funky just outside of warranty , most of the larger decoupling caps are Nichicon , then theres a few others of suspect branding , they look Chinese ,maybe thats the issue .
Its without doubt this unit is designed with the same attention to detail as the rest of the range , mic and line input still work fine in every respect .
I could take a chance and email SSL , theres no doubt the issue has arisen before , its documented , but monoliths have a way of burying their heads in the sand when they want to so you can end up talking to the wall .

 

[HP200CD]

Its the driver version 4.67 that works with Win7 ,
they dont make that clear on the website ,

Hi Tubetec,
Yes I am using driver version 4.67.0 with Win7 and it works well, at least for Zoom, as I have no DAW on this machine.

I've repeated the test with the headphone volume full
Let me confirm the other conditions:
Channel 2 (right) maximum gain, Hi-Z mode, open input, 4K off, headphone level full.
No oscillation, but lots of white noise and computer hash

 

[Tubetec]

If you make a connection from the audio input 0v , which is connected to chassis at the combo sockets ,to the output 0V those noise spurs go way down in level , circuit noise starts to become the dominant source . Im not sure if this would cause other unwanted interactions when you start hooking it upto other equipment though .