cc2590 and cc2500

Ryan,

The output of the CC2500 is 80+j74 and the differential input of the CC2590 is 105+j35.    You can see it will have a small mismatch.    The ouptut of the CC2590 is 33-j7 so it definitely requires a matching network to connect to most antennas.    The process of matching power amps to radios is more complicated than matching to passive devices.     Since the input matching can change the bias current which results is inproper matching of the output and stability issues and loss of output power. 

If you are not confrontable with doing differential matching then I would suggest using the recommended connection in the combo datasheet.  At worst you might not seet the 12 dBm gain of the part but should be stable for 50 ohm load. 

Rgds,

Thanks RRS,

By "combo datasheet", do you mean the CC2590 datasheet? The recommended connection according to this datasheet is a direct connection between the CC2500 and CC2590 with no matching components.

Another TI representative has given me the advice of using the CC2520->CC2591 reference design as a starting point. I might do this because it offers the flexibility of tuning the matching network in the future.

-Ryan

Using the CC2520-CC2591 connection would be a good choice since the output impedance is the same as the CC2500.   Work of caution the board thickness and or layer thickness can change the trace impedances so note the layer informaiton of the example you reference to and adjust appropaiately with PCB line calculators if you decide to change the stack up. 

Regards

Hi, i am also a desperate designer around these chips. I would ask the Ti staff who is tryng to help to be a bit clear, it is really ennoying to spend time around a critical aspect of the design, in my case with a fast deployment plan a confusion now will mean lots of work and money lost. We need either clear measurement data or a guaranteed interface, in my case CC2510 to CC2590. I found very relative suggestions for example ina  message above i found this:

"The output of the CC2500 is 80+j74 and the differential input of the CC2590 is 105+j35"

In the CC2500 and CC2510 data sheet, 80+j74 is the required load impedance to the CC2510 (in transmission). The CC2510 output impedance also during transmission is then 80-j74, if we assume the specification looks for conjugate impeance matching. So the sentence is not true concerning the CC2500.

Can I be sure that 105+j35 is the true CC2590 input impedance for transmission? Or the complex conjugate?

As to CC2520, somebody talks above about a "2520/2590 combo spec" I would like to know where is that. He says that 2520 impedance is the same as 2500, so also identical to the 2510. But this is not written anywhere in the data sheet. Can we trust?

Concenrning the 2590 I learn here that the input impedance (in transmission, output in reception I guess) is 105+j35 (or maybe its conjugate...) . But if so, this does not match very well with the application circuit which uses a 1 to 4 transformer to match the input to 50 ohms. Then the impedance seen at the RF input port is in the order of 25ohms much lower than 50 ohms.

Tthese 2.4GHz developments that can be destroyed if there is bad impedance matching. Can somebody really help with absolute data or recomendation on the 2510+2590.

Otherwise, I think the only safe is match perfectly the 2510 to 50 ohm with a commercial balun and then, add a 50 ohm to 105+j35 (if somebody can certify that is the CC2590 Tx input impedance) and get also guarantee that the Rx impedances are similar to the Tx ones.

Hi Jaime,

One of the problems here is that the RF outputs do not have one certain impedance. The RF ports are connected to the TX PA and the RX LNA which are both active devices and  will both have a certain range (or areas in the smith chart) where they are stable and work well. What we state in the datasheet is "optimum load impedance" which is what the differential impedance of the balun should be when looking into the balun. This value is derived through load pull measurements in the lab and is the best balance between TX- and RX performance. The same goes for the CC2590 inputs, there is not one certain impedance, but rather a range of impedances that will work well with the PA and LNA.

There are a couple of reference designs you can look at. First off we have the CC85xx boards where the CC2590 is connected directly to the CC85xx. These boards work very well. We also have a reference design for CC2530 with CC2591 along with an app. note describing measurements done on the same board with CC2590. This app. note is found here: www.ti.com/lit/ml/swra375/swra375.pdf. (Note that when replacing CC2591 with CC2590 theres a small change required for the decoupling also.)

We do not have a reference desing for CC2510 with CC2590, and we do not plan to test this combination either. My suggestion to you would be to make a prototype board based on the CC2530-CC2590 design and test the performance both with the matching network between the chips and without.

Best regards,

Fredrik

Thank you Fredrik. I will check the documents you refer to, specially the 2530/2591 one.

In any case, to have as reference a layout or a partlist is not a very secure procedure. The layout impact is large, the change of PCB leaves one lost, and one can't optimize variations. really what is necessary if the device parameters are complex, is to have a electrical definition of the networks (dipoles or quadripoles). In this way from your comments, i would think this way:

- for CC2510 direct to antenna I will use the Anaren balun which with the optimized circuit will present a impedance of 80+j74 at the CC2510 differential port

- for the CC2510 + CC2590 chain:

a) the CC2510/CC2590 interface can be designed as to present each device the impedance it likes, so suppossedly must present 80+j84 at CC2510 output and 105-j35 at CC2590 input (it is written here that the CC5590 input impedance is 105+j35 ???)

b) The output stages (filter, antenna) must present 33+j7 at the CC2590 output? because it is written somewhere here that its impedance is 33-j7 (???).

So for now, my plans would be

2510 alone: -CC2510-Anaren balun for CC2500-CC2510 recommended 50 ohm PI LPF- 50 ohm antenna

2510/2590 chain: three options,

a) CC2510-Anaren balun for CC2500-4:1 Murata transformer for CC2590- CC2590- recommended T matching network- 50 ohm Johanson ceramic LPF (or 50 ohm SAW filter)

b) CC2510-direct connection to CC2590-CC2590- CC2590 recommended T matching network- 50 ohm Johanson ceramic LPF (or 50 ohm SAW filter)

c) remove chip reactances and direct connection

I simulated the three assuming that the impedances one see at the chip ports are 80-j84 and 105+j35 with these results

a) in this solution i added an 1.8pF series capacitor at the CC2590 input (better 2 3.6pF balanced) to remove its reactance also, getting less that 0.7dB losses (but the Murata transformer losses add 1.4 so the total losses would be about 2.1dB), with good impedance matching better than 16dB. Of course it will have also sensitivity to traces but each trace segment is totally impedance defined so it will not affect mismatch.

I konw this is an expensive an artificious solution anyway... 3 baluns including the 2590 internal one.

b) WIth a ideal direct connection, the port impedances produce 0.3dB losses and 12dB return losses, but in this case the impedance is not totally controlled. It enworse fast when UIe add thin and long traces producing mismatch and up to 3dB and 3dB RL losses if the traces are too long (>5mm) and/or thin (<.5mm) for a 0.35 or 0.7 mm substrate. Nevertheless, the result keeps acceptability (0.6dB losses, 10dB RL) if one adds up to 6mm traces but taking care using an intermediate impedance (about 90-100ohm).

A variant of the first solution, with only one transformer, is adding -j35 at the 2590 input and +j74 at the CC2510 output i.e. 4.8nH and 1.9pF. It is better to use them in differential mode. Then the impedances between them become real, 80 and 105 ohm. Simulating a direct conection with this, the result for no traces is less than 0.1dB losses and RL 17dB. Adding the impact of intermediate traces with the geometric impedance media 91.7 ohm the impact of a short trace (up to 5mm) is lower than 0.6dB losses and 10dB RL, a result the same good than the direct connection but with the flexibility to be able to modify a bit the L and C values if needed.

A small interstage mismatch loss is not critical for transmission as far as the maximum power required can be reached, but I ignore the effects on the power amplifier behaviour. Concerning reception, as we add the 2590 only in the fixed base readers, the LNA role is less important than the HPA, that helps the tags receive better. So I guess the main thing to take care is avoid large mismatch losses in the interstage.

I would like to know your opinion about these circuits, thanks agan

JM

Another independent and i hope easier question (I am sorry i can't find where to start independent threats in this hug web). After reading the CC2590 data sheet, I thought that the recommended CC2510/2590 control interfaces are GD0 to PAEN and and GD2 to EN, and HGM not coming from 2510 but from other controller or tied to fixed value.

My interpretation is that I should use some of the P1_5,P1_6 and P_17 CC2510 pins that can be configured as GD0,1 or 2. I don't find somethiong very clear in the 2510 data sheet (at least for one who never used this kind of devices) but after reading, I connected a line P1_5 to PAEN and another from P1_7 to EN (LNAEN), because P1_6 seems to have other uses,a dn trying to guess why the 2590 requires in particular GDO0 and 2.

Concerning HGM, I left it tied for maximum gain.

But I am redesigning previous breadboards and i see that in the previous application, the interface was made this way:

PAEN from CC2510 P2_0

EN from Cc2510 P0_7

HGM control from CC2510 P0_6

I am not sure if i understood well the documents, it lokks as if it contradicts the recomendded scheme in taking HGM from the 2510, apart from that, would like to have a bit light about what are the CC2510 GDOs, and pins used for them.

Jaime

Hi Jaime,

We consider having a recommended layout and partlist a safer reference than just giving out the impedance values. Of course a requirement for this to work is that the designer follows our design as closely as possible. The reason behind this is that many regular EE are not used to RF design, and simply calculating component values in the Smith Chart will often give you incorrect results as lumped components have paracitic inductance and capacitance. Also paracitics in the PCB will not be accounted for. Our reference designs are based on simulations and lots of tuning in the lab, so by following these designs, the path to good performance will be short (some tuning will almost always be necessary).

- for CC2510 direct to antenna I will use the Anaren balun which with the optimized circuit will present a impedance of 80+j74 at the CC2510 differential port

Yes, that will be a good solution. Another option is to follow our reference design with passive components.

- for the CC2510 + CC2590 chain:

a) the CC2510/CC2590 interface can be designed as to present each device the impedance it likes, so suppossedly must present 80+j84 at CC2510 output and 105-j35 at CC2590 input (it is written here that the CC5590 input impedance is 105+j35 ???)

b) The output stages (filter, antenna) must present 33+j7 at the CC2590 output? because it is written somewhere here that its impedance is 33-j7 (???).

Yes, this seems correct.

2510/2590 chain: three options,

a) CC2510-Anaren balun for CC2500-4:1 Murata transformer for CC2590- CC2590- recommended T matching network- 50 ohm Johanson ceramic LPF (or 50 ohm SAW filter)

b) CC2510-direct connection to CC2590-CC2590- CC2590 recommended T matching network- 50 ohm Johanson ceramic LPF (or 50 ohm SAW filter)

c) remove chip reactances and direct connection

I would definately go with solution C here. This also gives you the option to experiment with direct connection.

For connection between CC2590 and antenna I would also recommend the passive component solution used in the datasheet I linked to in my previous post.

BR,

Fredrik

GD0 and GD2 are name of GPIOs on CC2500.

For CC2510 you are free to use whichever GPIOs you want, and you can also connect the HGM pin to a GPIO if you want the possibility to actively change between high- and low gain mode.

/F

Hey Fredrik I asked this in Charlotte thread but no reply can you assist here please. Some changes were decided in this project, one of them is to discard the "powerful" version using the CC2590. All the tags and readers will be based on CC2510 alone. But I wanted to ask you something else, talking with our partners, I suggested that maybe the CC2530 or other members of the chip family are more suitable for an RFID or RF ID/LOC in an interferred environment than the CC2510. But I don't have access to the technical parameters of quality and traffic of our customer to evaluate the needed BER and bitrate. So i just suggested them that DSSS chips like CC2530 can be better because the channel is notably more robust with DSSS and CSMA, and if they are using this system to implement some kind of persons or items presence control or data gathering in an open environment where dozens of WiFi or Zigbee or other more robust systems are working i just figured out that the CC2510 is in inferiority, unless they use very wisely the frequency hopping. Certainly, this can be convenient only for high data rate, not to compare the CC2510 at 2.4kbps with the CC2530 at 250kbps. i am not sure because i don't know how much cochannel interference the CC2510 receiver affords, but without FH the system maybe will be blocked below a positive C/I i.e. with interferences lower than the signal. but since Can you give us some idea fof comparison that I can translate to our partners in order to asses them to choose well?

Thanks in advance

JM

Hello I posted previously when designing the RF circuits. Now we have our own CC2510 board and we have respected the debug intrface literally for accessing using the CC debugger flash programmer and Smart RF Studio (release 7). We can successfully program the chip using Flash programmer (the only problem was that we can't do it in the "fast" mode but it succeed in the "slow" mode. After that (and before) i have tried repeatedly to program the CC2510 RF functions. Our system works perfectly with TI Smart Cc2510 RF EB's (we can transmit, receive, and all Rf behaviour is correct), but when using our board instead something very strange happens: one can transmit continuously or in packets in the Easy and the Expert modes but the RF signal is not correct. It transmit something, it is a carrier in the correct frequency band and we can control perfectly the output power, either in continuous mode or in packets, but: 1) the signal is never modulated (suppossedly the test modulates it with nternal random data); 2) the signal is not exactly in the selected frequency: normally it is about 15-16 MHz below, the carrier quality is not good and the frequency is not completely estable. My impression is that the PLL synthesizer remains unlock, as if the VCO bank was not calibrated or so. there is nothing in the data sheet explaining PLL lock issues and calibrating procedures, but the TI boards work correctly even so, and we can't get it with ours. Our RF circuit is completely tested and the performance is very good; also i ckeched the crystal oscillator and it looks correct (when you open the Smart RF Studio, a 26MHz clock signal appears with correct amplitude (1.6Vpp, the same as in TI boards), we measured the frequency carefully using resistive coupling not to pull the frequency and the result is well into specs (less than 10ppm error). This happens when transmitting: of course i can detect the power with a Smart EB TI receiver, tuning it in the frequency that i measure from our board, but only detect power, no data as the signal is not modulated. the levels are correct anyway! The reverse happens when we receive, even if this is more difficult since we restart the chip using the "Continuous Rx" mode for example and sometimes the chip changes the frequency, but even so I have been able to detect the EB Tx frequency placing its frequency where I believe to be our board CC2510. The received power is also correct and reacts exactly to changes. I believe all this is a problem with the PLL and modulator programming, due in origin to the fact that the synthesizer never reaches the correct frequency, and this may be due to some autocal procedure that fails (???). I tried to "manual cal" once but the effect was crazy, the tx signal in continuous mode started to move upwards in frequency towards the correct frequency but before reaching it the signal dissappeared, and this behaviour became permanent even switching off, until i left desperate home; the next day all was again as in the beginning. This behaviour is what makes me believe that we hae an issue with the synthesizer, but i can't understand how this can be related to our HW. I can send you all the details you can need about our HW and the test bank, including the register values. We have a problem with this chip becaise compared with any other we use, the documentation about its internal RF functionning is really short, and there is near anything about the synthesizer and the modulator. I have checked that we respect literally the debugger interface that is documented to be the only thing the Studio needs to work properly. Please give us a path to debug this problem.

Thanks in advance

Jaime Martin

Hi Jamie,

This may very well be HW related. Could I have a look at your schematic and Gerber files?

Rgds,

Fredrik

_{Thank you Fredrik, sure we can send you the schematic first, then we could check the gerber if you think it can affect.}

_{Please tell me how to send you a PDF with the schematic in first instance (to an email address?) because it is not allowed to me to make it public since this is a design for thirds. Can you tell me also what you mean HW related, I discarded the RF section because all the S parameters are checked (Tx output, Rx input, antenna matching, Tx losses), also I discarded the clock circuits since the performance is good (less that 10ppm), I checked also the power supply in DC and AC and I didn't check yet line by line but reviewed the compatibility of the HW interface. The supply has a particularity that this is a battery circuit but we toke care that the debugger/circuit power supply are compatible in the tests using external power supply.}

_Jaime

Hi Jamie,

Please send the schematic to.

/Fredrik

Thank you Fredrik, I am preparing them now.

A good news, i have finally discarded the whole power supply circuit, not only the DC/DC as before: now with a simple power supply 2.0 to 3.6V as we foresee the whole radio is working perfectly.

I dont know exactly the reason why it does not work with the battery, does not matter if it is perfectly charged, or even while recharging from the USB. Does not matter if the voltage is low or high, the failure was the same: no LO frequency lock and no Tx modulation. The only hint i have is that unless the battery is very well charged, when transmitting the chip AVDD voltage from the battery drops continuously but no more than some dozens millivolts per minute

Jaime,

Please keep me posted when you figure out the problem with the liion charger and/or DC/DC regulator.

/Fredrik

Sure I'll do Fredrik. We start now to prepare a new bank to verify all the power supply issues.

In the meanwhile if you are so kind to tell me if the quality figures I measure with the analiser (frequency accuracy, phase noise, spurious in and out of band, etc.) are OK

Jaime

Jaime,

Your measurements look good!

/Fredrik

Thank you Fredrik your comments are very valuable. I have measured also the ripple in band and the output circuit losses and thet arevery good, the first is about 1 dB slope from 2.4 to 2.4835GHz and the second is a few tenths of dB including our small Rf connector, the miniature panelable cable and the spectrum analyzer main cable. I have to recommend the JTI balun that we use for new users because it is the key that the output circuit behaves so well. I have also made recordings in all the Smart RF Studio modes from narrow band BFSK to widest MSK all with PER=0 both in Tx and Rx.

We are still investigating the reasons the board didn't work at the beginning, because once we have been able to load our program in flash, it worked inmediately, both with the external supply and with the original infraestructure. the DC/Dc is abe to deliver 100mA so i don't atribute the failure to its current capacity unless the debugger would need near 60mA which i don't believe... maybe it is the way the CC2510 start to be loaded which is difficult for the DC/DC ??? I will tell you when we guess, because the strangest thing is that after all this happenned, we have deleted all the flash and start from scratch with the deleted 2510, the battery, the DC/DC, the debugger, the Flash programmer and the Smart Studio and all works finely from the beginning. It is "as if" the CC2510 "behaves better" from the moment it was loaded for first time with our program in flash (because our program, oppsite to the smart RF studio, program the 2510 P0_0 output depending on its state "sleep" or active. Remember that our HW following TI appnote about DCDC includes the debugger/smart studio minimum HW but includes also a "sleep" mode that make the things more difficult in the very beginning (we call in spanish a fish that eats its own tail)...the final, real problem is that, even if the 2510 tells the DCDC to bypass (when the former is sleeping), the voltage is still OK both for 2510 and for debugger (2.0V when active, 3.6V when sleeping) so i figure out that all should work, always, with no special caution or jumper for the initial flash recording... for now that is all...