[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[linrad] Presonus Firebox Help Needed!!

HI, All,

This is a long post and so if you are not interested in helping me with a problem I am having with the Presonus Firebox, delete NOW.

This is another post regarding problems I am seeing with the Presonus Firebox. I am hoping for one of the I/Q gurus to help me figure out what is going on. My questions are basically:

1.  What is happening?
2.  Is it a hardware or software problem?
3. Is this behaviour present in all Fireboxes, in which case I will keep it, or is it unique to mine in which case I will send it to the manufacturer under warranty.

I am seeing MAJOR problems with the I/Q signals sent by the Firebox over the Firewire to the computer when it is operated at 96kHz sampling rate. These problems are not present at 48 kHz sampling rate, so I do not think they are a just a result of improper soundcard connections by me. As I will show below, they are NOT the expected 'image' problems seen when one of the I/Q signals is not present, or is out of phase or of inappropriate amplitude.

I am hoping that someone on one of these lists can explain to me what might be going on.

I would also be interested in whether anyone else is using the FIrebox at 96 kHz with Linrad or Winrad [or other software that attempts to see an approximately 990 KHz slice of the RF spectrum]. If so I would be interested to know if when they have systematically looked for this problem they have seen similar things.

The problem [since it is not an image problem] is of course not corrected using the Channel Skew correction available in Winrad.

The problem is present in both Winrad and Linrad when using the Presonus Firebox as the input soundcard when sampling at 96 kHz, I/Q mode receive, with the SoftRock v6 for 80/40 meters being used for input to the Firebox. It does not occur at all with the FIrebox run at 48 kHz, and NEVER occurs when the Delta44 is used as input soundcard at 48 or 96 kHz with the SoftRock as input. I am quite confident that the SoftRock is working appropriately, as I have had no problems with it when using the Delta44 with it.

Here is the data:

If I disconnect one of the I/Q signals coming into the Firebox from the SoftRock I get just what one would expect: The signals are mirrored about the center frequency and sidebands are inverted on the image signal. This is normal to see. This is NOT what I see when both I and Q inputs are connected to the Firebox [one to Main-In 3 and one to Main-In 4, each using a TS 1/4 inch Phone Plug]. I did this experiment just to prove that the problem I am describing is NOT an image problem. I knew this, but was certain that some reading this wouldn't be so sure.

The malfunction is this:

There are identical spectra displayed above and below the center frequency. I used the signal from CHU at 7.335 MHz as my 'signal generator' for these tests, as all of my test equipment is at a remote site. The CHU signal can be detected between the center frequency and up to 23000 Hz above or below center frequency, but not beyond [i.e., the bandwidth over which an input signal can be detected is about 48 kHz or half the sampling rate, just as if the card were running at 48 kHz or if I/Q sampling were NOT being performed]. But there is noisefloor displayed over the full 96000 bandwidth of Linrad or Winrad, and the CHU signal will be displayed over the full 96000 bandwidth as one varies the LO frequency due to the 'extra' signal appearing on the other side of the spectrum from the real signal, as is shown below. The displayed frequency for signal appearing on the Linrad display within 23000 Hz of center frequency is 'appropriate'. The frequency of the 'extra' signal is of course not appropriate but varies in a predictable way. The table below gives the LO frequency and then the frequency of the 'correct' Linrad CHU signal and of the 'extra' Linrad CHU signal appearing on the opposite side of center frequency in the Linrad display. The behaviour is the same with Winrad although the frequencies are labeled differently in that program. Note that the spectra are NOT mirror images of each other and there is NO inversion of sidebands [if you can't see the latter point, take my word for it. I have confirmed it both visually and audibly]. This problem is NOT an due to an absence of one of the two I/Q signals. I think that the Firebox Driver must be handling the I/Q signals incorrectly at 96 KHz. But the explanation is what I want an opinion on. Here is the data table:

LO          Relative Freq as
FREQ     Displayed by Linrad
IN Hz      With Firebox input
              [Correct /  Extra]

29250     No CHU signal
29250     22500  74500
29260     20000  72000
29270     17500  69500
29280     15000  67000
29290     12500  64500
29300     10000  62000
29310     07500  59500
29320     05000  57000
29330     02500  54500
29340     00000  52000
29350     97500  45500
29360     95000  43000
29370     92500  40550
29380     90000  38000
29390     87500  35500
29400     85000  33000
29410     82500  30500
29420     80000  28000
29430     77500  25500
29440     75000  23000
29450      No CHU signal

WIth proper function, there would be CHU signal down to LO of 29160 and up to LO of 29530. This is the case when the Delta44 is used as input.

The Linrad display is set up so that the bottom frequency is approximately 52500 and the frequency then increases to 99999 just below center frequency, which is 00000. The frequency just above center frequency is 00001 and frequency continues to increase up to a maximum of about 45500. So frequencies between 52500 and 99999 are below center frequency, and frequencies from 00001 thru 45500 are above center frequency. So the CHU frequency starts out near the middle of the upper [right] side of the spectrum and gradually decreases or moves to the left until it disappears at about the middle of the lower [left] half of the Linrad display. Doing the experiment in Winrad gives the same result.

I have placed gif pictures of the Linrad display with the FIrebox operating in this way and the LO at 29440 Hz at


You can see that the two sides of the spectrum appear identical [except for the gradual shading caused by the very non-flat frequency response of the Firebox]. You can see the 'correct' CHU signal at 75000 on the left side of the screen, and the 'extra' CHU signal at 23000 on the right side of the screen. Note that as I indicated above, the signals are NOT mirrored as would occur with an imaging problem.

For comparison I placed a gif picture of the Linrad display with all conditions identical, but with the Delta44 sampling at 96kHz used as input instead of the FIrebox at


You can see that the CHU signal at 75000 is NOT reproduced at 23000 on the right hand side of the spectrum. There IS a much weaker signal at 23000 to 25000 and that IS the weak image signal of CHU, and it does have inverted sidebands as expected. This is a normal result of imperfect I/Q phasing and does not indicate a problem at all. All of these tests were done without I/Q correction for obvious reasons. Incidentally note the MUCH flatter frequency response and thus flatter noise floor with the Delta 44 as compared with the Firebox.

These experiments show that the problem is not with Linrad [or Winrad; data not shown] or with the SoftRock.

I can change the sampling rate Linrad thinks it is seeing, and when I change that to 48 kHz I get entirely different results. If I change the sampling rate of the Firebox to 48 kHz I get correct function of the FIrebox and thus different results.

I hope someone out there can explain this to me.

For the cognoscenti, I have placed Linrad "S" files and wav files of the 40 meter CW contest last night that were made using both the Firebox and Delta44 as inputs, thinking that this might provide additional information. I will leave those files on the server for only a couple of days, as the space they occupy is generously loaned to me and I do not want to abuse the gift with these large files. They are 12-15 MB size.

Firebox S file
FIrebox wav file
Delta44 S file
Delta 44 wav file

I also posted gif files of the spectral analysis done by CoolEdit96 of these wave files. It looks to me like the Firebox spectra are mirrored around center frequency rather than showing unique information on either side of zero frequency. I suspect that is the root of the problem, but don't know why it is occurring. Maybe the Firebox actually only looks at 48 kHz of bandwidth but oversamples at 96 kHz?? Would that give a useful input bandwidth of 48 kHz but an output noise floor that covers 96 kHz when the card is set to sample at 96 kHz?

Those files are at:
http://home.comcast.net/~w3sz/fireboxwavespectrum.gif for the Firebox wave file http://home.comcast.net/~w3sz/delta44wavespectrum.gif for the Delta44 for the delta44 wave file

Thanks in advance, and


Roger Rehr

Roger Rehr

This message is sent to you because you are subscribed to
 the mailing list <linrad@xxxxxxxxxxxxxxxxxxxxx>.
To unsubscribe, E-mail to: <linrad-off@xxxxxxxxxxxxxxxxxxxxx>
To switch to the DIGEST mode, E-mail to <linrad-digest@xxxxxxxxxxxxxxxxxxxxx>
To switch to the INDEX mode, E-mail to <linrad-index@xxxxxxxxxxxxxxxxxxxxx>
Send administrative queries to  <linrad-request@xxxxxxxxxxxxxxxxxxxxx>