linrad support: Testing and tuning the RX2500.
(Sept 07 2003)

Smoke test

Connect DC power to the 9-pin d-sub connector.

pin 9 = DC ground
pin 5 = +15V DC, 0.6A
pin 4 = -15V DC, 0.6A

In case the current differs much from the nominal 0.5A something is wrong. There is not yet any experience on faulty boards so I can not give detailed hints for troubleshooting. The total current is used in three roughly equal parts. 0.15A for each of the RF amplifiers and 0.2A for all the 24 AD979 op-amps. The current through the RF amplifiers is conveniently checked by the voltage across the 22 ohm resistors in their power supply lines. It may be a good idea to have current limited power supplies to avoid burning the 1A fuses. In case some error causes an RF mosfet to have near zero resistance, the power in the 22 ohm resistors could become 8W and since these resistors are rated 3W only they may become damaged if the fail condition lasts for many minutes.

Coarse tuning of the RF filters

Connect a signal generator to both RF inputs with a T-connector. Set the frequency a few kHz from 2.5MHz. Anything between 1kHz and 25 khz will be fine. A suitable power level is -20dBm, but the level is not critical.

Select I=SOUNDCARD TEST MODE on the main menu to display the audio signal with the oscilloscope mode. (In early versions of Linrad this function was called F=HARDWARE TEST MODE.) The keys "1" and "2" are used to select the channel and "+" and "-" can be used to set the oscilloscope gain.

If there is no signal, the X-tal oscillator may not be running. Set C155 for maximum capacitance and tune C157 until you see the signal.

Reduce C155 until the oscillator stops and fine tune C157. The correct tuning of C157 is when the oscillator runs (starts oscillate when power is applied) with the smallest setting of C155.

When C157 is correctly set, tune C155 for the correct frequency. Use a frequency counter or apply a signal of known frequency and tune for the normal receive mode to place the signal at the correct frequency.

When the X-tal oscillator runs properly and a signal is seen in the oscilloscope modes, tune the capacitors C99, C104 and C105 for maximum signal in channel 1 and C158, C148 and C149 for maximum signal in channel 2.

Check the gain margin of the X-tal oscillator by connecting a 3.3 kiloohm resistor across L34. The oscillator should start nicely with this extra load on the collector side of the transistor. In the first batch of 25 units, 3 did not pass this test. Two X-tals had too high series resonance and one transistor insufficient gain.

Tuning the audio notch filters

Change the frequency of the signal generator for the audio frequency to become 49.2kHz. One way of doing this is to use the normal receive mode and tune the signal generator for the signal to go below 0 kHz. The signal is then aliased back into the passband and should be set to 1.2kHz. Expand the spectrum to see the frequency scale well.

Tune L3 and L16 for minimum signals in channel 1 and L1 and L6 for minimum signals in channel 2.

In case an oscilloscope is used, measure the output voltage at these points:
R76 for L6
R91 for L1
R110 for L3
R115 for L16

Then change the signal generator for the audio frequency to become 55.0kHz. The Delta44 is very insensitive here due to the builtin anti alias filter, but it is usually possible to follow the alias signal up to 7.0 kHz if the signal generator power is raised to 0dBm.

To tune the notches at 55kHz an oscilloscope or some other means of measuring audio signals is needed. For 0dBm at 2.445 the output is typically 100mV when the notch is tuned for minimum.

Tune for minimum. Measure the output voltage at these points:
R76 for L5
R91 for L2
R110 for L4
R115 for L15

Fine tuning the RF filters

Connect a pulse generator to the input of channel 1. Make sure that Linrad is uncalibrated by removing all files that start by dsp_

For tuning a 5V p-p square wave with a repetition rate of 100Hz is suitable. When the unit is only coarse tuned as described above, the screen looks typically as shown in fig. 1.

Fig. 1. Typical frequency response of coarse tuned unit.
Fine tune C99, C104 and C105 for a symmetric spectrum with as much amplitude as possible. A typical result is shown in fig.2.

Fig. 2. Typical frequency response of fine tuned unit.
The frequency response is not very important. The Linrad calibration procedure corrects for all filter errors and the only consequenses of a non-flat frequency response is changed dynamic range properties.

As the unit is designed now, gain increases slightly at high audio frequencies. This is to partly compensate for the higher noise floor of the Delta44 at high audio frequencies.

One could also allow the amplitude to fall by up to 10 dB towards the spectrum ends to improve the tolerance for undesired signals but then the system noise figure would be degraded for higher audio frequencies.

The frequency response is easy to change, but at the present time I do not know in what direction it would be desireable to change the dynamic range properties. The present solution is sort of middle way.

Checking RF linearity

Connect two signals, both with a power level of +10dBm to the input. Make sure the IM3 level of this test signal is below -50 dB regardless of the load impedance. Check the levels of IM3 products at the output of the RF filters. Use a high impedance probe at the output side of L24 for channel 1 and L31 for channel 2. The IM3 products should be 35dB below the main signals. For the first 25 units, the IM3 level was between -34.8 and -37.4dBc corresponding to an input IP3 of +27dBm for signals within the visible passband. For signals outside the RF filter, IP3 is higher, about +32dBm.

Mixer and AF linearity and balance

The Delta44 does not quite have the linearity to test the 2.5MHz to audio converter. To check for second order distorsion, apply an RF signal that gives a signal level about 1dB from saturation of the Delta 44. Then attenuate the audio signals 10dB with suitable resistors (a 10dB four-channel audio attenuator with 15-pin d-sub connectors is handy here)

Second order distorsion should be at least 100 dB below the main signal when the main signal is placed at about 10kHz. While checking the first 25 units, one was found with the second harmonic 97 dB below the fundamental. This was cured by replacing one of the 74HC4052 IC's.

While checking the level of the second order harmonic it is convenient to check for the balance between I and Q. This is done by checking the level of the mirror image. It should be 35dB below the main signal or lower.

The mirror image should be checked at a higher frequency also since there are many components in the anti-alias filter and it is perhaps possible to have errors that only affect the gain well above 10kHz. Inject a signal at about 2.540 MHz and compare the main signal at 88kHz to the image at 8kHz. The difference should be at least 20dB.

Checking the noise floor

The Delta 44 is normally operated in is lowest sensitivity mode ("+4dB" with ossmix) to minimize the noise contribution from the internal amplifier in the A/D converter. To check the noise floor of RX2500, set the Delta 44 gain to "-10dB" which will make the gain 12.0 dB higher while the noise floor is lifted by 3.6dB. The sensitivity of Linrad will be as it would have been with an A/D converter having 8.4 dB higher dynamic range but saturation will happen 12dB earlier. The boards are checked by use of the average reading of the S-meter fort H-pol and V-pol at a frequency of 65 kHz. Both channels on all boards give the same noise level, 3.3dB above the Delta 44 at "-10dB". At the normal gain setting of the Delta 44, the 2.5 MHz to audio converter lifts the noise floor by about 0.8dB.

Checking close range reciprocal mixing and noise modulation

A test signal at 2.510 MHz, 1 dB below saturation is injected to the channels one by one. The noise floor at a frequency separation of 5 kHz should not increase by more than 2dB in a single channel.

No errors were found with this test All units behave identical, possibly because the test signal itself is the limiting factor.

Checking medium range reciprocal mixing and noise modulation

A test signal at 2.549 MHz with a level of about 0dBm is injected via a notch filter at 2.524 MHz. This test signal will be stopped by the first notch of the anti-alias filter so very little signal will reach the Delta 44 which is run in high gain mode ("-10dB"). The mixer will see a very strong signal, close to cause saturation of the second AD797.

The noise floor should not rise by more than 1.5dB.

This test gives the noise floor at 25kHz separation for signals outside the visible passband. The test signal is 15 dB above the normal saturation limit for the Delta 44 but since the card is run in high gain mode, the test signal is actually 27dB above saturation during this test. The noise floor at 72kHz, the frequency for the test is at -141.5dBc/Hz with respect to a signal that precisely saturates the A/D converter in high gain mode so the noise floor of the entire 2.5 MHz receiver is at -168.5dBc/Hz with respect to the test signal. Since the noise is lifted by 1.5dB (or less) the contribution due to reciprocal mixing and noise modulation is about 4dB below the noise floor of the 2.5 MHz receiver itself which means that this test ensures that the noise floor is at least as good as -172 dBc/Hc at a frequency separation of 25kHz.

In the first test of 25 units, four of them failed this test, giving a noise floor degradation up to 2.5 dB. These units became normal with a replacement of the 74HC4052 mixer IC's. It seems like the dominating source of phase noise is these IC's. It is possible to bring the noise floor down to -174 dBc/Hz by selecting particularly good IC's but there is no reason. It may be ok to allow 3dB noise degradation (-168dBc/Hz) to avoid having to replace any IC's. The performance of the other converters intended for use together with RX2500 might make this level fully adequate.

Checking sensitivity

Set the frequency of a signal generator a few kHz away from 2.5MHz. Connect the generator to channel 1 and set Linrad in normal receive mode. Press "A" to get amplitude margins on the screen.

The level required for saturation should be about -15dBm. The amplitude margin readout holds the peak value of the signal. Press "Z" to clear. The amplitude margin readout should show by how many dB your test signal is below the saturation limit.

Move the signal source to channel 2 and check that the margin to saturation is the same within 1dB.

The first batch of RX2500 was tested with a test signal at 2.508 MHz at at level of -30 dBm (+/-0.1dB). The amplitude margins are shown i table 1.

Board      AD1       AD2       AD3      AD4
 No        (dB)      (dB)      (dB)     (dB)    
  1       17.72     17.64     17.54     17.53
  2       17.65     17.56     17.59     17.60
  3       17.58     17.45     17.52     17.52
  4       17.61     17.61     17.51     17.48
  5       17.66     17.52     17.46     17.49
  6       17.57     17.46     17.56     17.55
  7       17.82     17.76     17.62     17.65
  8       17.60     17.44     17.53     17.53
  9       17.52     17.49     17.50     17.61
 10       17.61     17.56     17.56     17.59
 11       17.69     17.63     17.43     17.49
 12       17.64     17.50     17.54     17.61
 13       17.66     17.56     17.52     17.55
 14       17.64     17.59     17.52     17.53
 15       17.58     17.41     17.48     17.52
 16       17.65     17.58     17.71     17.69
 17       17.70     17.63     17.55     17.65
 18       17.52     17.42     17.52     17.43
 19       17.61     17.53     17.40     17.49
 20       17.63     17.56     17.52     17.51
 21       17.78     17.74     17.60     17.61
 22       17.70     17.63     17.55     17.57
 23       17.57     17.55     17.60     17.58
 24       17.67     17.63     17.61     17.63
Table 1.Amplitude margins for RX2500 with -30dBm at 2.508 MHz at the input. All channels are equal within a few tenths of a dB and the level at which the Delta 44 saturates is -12.4dBm.