Review: ICOM IC-9700 VHF/UHF transceiver
Sam Jewell G4DDK has the lowdown on the much-anticipated IC-9700 VHF/UHF transceiver from ICOM.
Like many others I have waited patiently for my IC-9700 to arrive, having finally decided to order one in mid-2018. I bought an IC-7300 three years ago and was delighted at its performance, ease of use and the convenience of the single USB connection for PC control. My hopes were high for the IC-9700.
The IC-9700 has many of the features of the IC-7300 but in a 3-band VHF/UHF transceiver. As well as the colour touchscreen display and a very comprehensive settings menus, the transceiver also has real-time spectrum scope and a waterfall display. The transceiver covers the 2m, 70cm and 23cm bands as standard. It also includes Satellite mode, with any combination of two of the three bands. It features three separate antenna connectors: an SO239 socket for 2m, and separate N-type connectors for 70 and 23cm. It also includes D-STAR digital voice (DV) on all three bands and the higher speed Digital Data (DD) mode on 23cm.
The IC-9700 is described as a VHF/UHF Software Defined Radio (SDR) using direct sampling. This is not completely correct, however. In order to cover the 23cm (1296MHz) band the architecture incorporates a conventional heterodyne conversion with a 310 to 370MHz first IF. This IF is then direct sampled, like the 2m (144MHz) and 70cm (430MHz) bands. Direct sampling avoids the limitations of mixing non-linearities and the need for crystal roofing filters which, in turn, can introduce other performance impairments. The IC-9700’s SDR architecture allows a great many other advanced facilities to be incorporated, including wideband, real-time, spectrum and waterfall displays, an extensive range of bandwidth resolution filters and expanded computer control. One of the most anticipated performance improvements, over conventional superhet (superheterodyne) based transceivers, was the potential to significantly reduce the noise added to both transmit and receive signals by the often-noisy conversion oscillators in superhet-based receivers and transceivers.
VHF and UHF enthusiasts have waited a long time for a new, modern, dedicated transceiver to appear. Recent introductions have generally been of the ‘shack in a box’ type, with HF and VHF/UHF coverage and often incorporating receiver general coverage of the complete HF bands and selected portions of VHF and UHF. Of necessity these rigs often use superhet up-conversion to a first IF in the VHF range. The then-necessary VHF first IF crystal roofing filter can exhibit limited performance, putting the remainder of the receiver under greater performance constraints. The true SDR doesn’t need this roofing filter.
Does the IC-9700 live up to the hopes and expectation of the many VHF and UHF amateur radio enthusiasts who have bought or plan to buy one? Read on to discover if this is the radio you have been waiting for.
The IC-9700 is housed in the same compact case as the IC-7300. The front panel features a 4.3in TFT LCD colour touchscreen with several ‘soft’ buttons controlling various menu selections. These are complemented by several encoder rotary controls and pushbuttons as well as the microphone, headphone and SD Card reader/writer sockets.
At first sight it looks identical to the IC-7300 but closer examination shows that there are two receiver controls, one each for the main and secondary receiver; slightly different labelling on the press-to-select buttons at the top right-hand side of the panel and instead of saying ‘D-STAR’ below the screen, as on the American version, the UK (and probably EU) version has the word ‘Digital’.
On the rear panel, Fig. 1, are three separate coaxial antenna sockets, a 10MHz reference frequency input socket; an 8-pin, full-size accessory socket; USB-B socket; a LAN (Ethernet) socket, power socket (13.8V) and various loudspeaker and keyer sockets. The rear panel is dominated by a rather large cooling fan.
Fig. 1: Rear panel view (from brochure)
While the IC-9700 can be used as a single band transceiver on any of the three bands, it can also be used single band but with simultaneous independent receiver (Dual Watch) on either of the other bands. But note, it cannot be used with both receivers on the same band. As well as the normal band selection options, the IC-9700 can be used as a full duplex (transmit and receive simultaneously) transceiver on satellite mode. Offset (conversion frequency) and normal or reverse tracking of the satellite frequencies is catered for in this mode.
The LCD screen can be set to show one band only or both bands simultaneously. Frequency, spectrum display and waterfall, together with receiver S-meter and relative power output meter can also be displayed. In addition, the status of parameters such as AGC, noise blanker, receiver filter, VFO A or B and Passband-Tuning, as selected, is shown. The time, in local or GMT, is also displayed.
ICOM claim 100W of transmit RF is available on 144MHz, while the power on 432MHz is 75W and just 10W on 1296MHz. A Japanese friend told me that the most popular Japanese licence type requires that the 1296MHz output is limited to 10W. These power levels can be adjusted independently on each of the three bands to suit the operator and mode. A useful facility to limit power to a set value is also incorporated. Reports indicate that there is no power spike, unlike many other transceivers that use ALC (automatic level control) to control output power.
As well as the usual CW, FM, SSB (USB and LSB), and data modes such as FT8, the IC-9700 provides for D-STAR voice on any of the three bands plus D-STAR data at 128kb/s on the 1296MHz band.
Inside the IC-9700
The basic architecture of the IC-9700 is shown in Fig. 2.
Fig. 2: IC-9700 internal architecture.
In a direct sampling receiver, a high-speed ADC (Analogue to Digital Converter) is clocked (sampled) at a rate at least twice the highest frequency of its coverage. To cover up to 52MHz, the sampling oscillator would need to operate in excess of 104MHz (commonly around 122MHz). Since the IC-9700 covers up to 1300MHz, you might expect the sampling clock to run at a frequency in excess of 2600MHz. Not so.
The sampling frequency, known as the Nyquist frequency, requires that the sampling clock only needs to operate in excess of the highest frequency encountered in the bandwidth of the signal of interest, not the whole bandwidth from DC to maximum frequency.
The US 70cm band, as an example, extends from 420MHz to 450MHz. That is a bandwidth of 30MHz. The sampling clock only needs to operate in excess of 60MHz to meet the Nyquist requirement. Provided the ADC input can process the highest RF frequencies encountered, the sampling clock can operate at moderate frequencies. In the case of the IC-9700 the sampling clock operates at 196.6MHz. This process of sampling the bandwidth rather than the whole DC to maximum signal frequency range is known as bandpass sampling or under sampling. Since bandpass sampling introduces a problem known as aliasing, good quality ‘anti-aliasing’ bandpass filters are used to closely define the band to be sampled. Increasing the frequency of the sampling clock to well above the required Nyquist frequency can ease some of the filter requirements. This filtering limits the radio to operating just within the amateur bands. No extended receive is available.
In the IC-9700 the 23cm amateur band covers from 1240 to 1300MHz. That is 60MHz bandwidth. Logically that would require a sampling clock frequency of over 120MHz. In practice the ADC used by ICOM has a bandwidth extending to 1250MHz, but that is not enough to be used (reliably) to operate to 1300MHz. The IC-9700 solution is to convert the whole 23cm band down to a ‘tunable IF’ of approximately 311 to 371MHz and then bandpass-sample that 60MHz wide band as you would the 70cm band.
The IC-9700 uses an LTC2156-14 dual-channel, 14-bit, 210MHz (maximum) sampling ADC with a full power bandwidth of 1250MHz. Dual-channel enables the two receivers to operate simultaneously.
Unlike the conventional superhet architecture used in most receivers and transceivers, the direct sampling architecture doesn’t need a VFO to select the operating frequency. The ADC (on receive) sampling clock is, in effect, the VFO, but has the advantage that its frequency doesn’t need to change with tuned frequency. In a superhet radio synthesised VFOs are usually the major source of unwanted noise in the radio path, apart from the inherent noise of the receiver early stages and the antenna noise. The noise on the VFO is a mixture of phase noise and amplitude noise, with phase noise usually dominating. It is introduced onto each received (and transmitted, in the transmitter case) signal and causes an effect known as reciprocal mixing. This is the effect you notice when tuning close to another signal and start to hear modulation from that signal due to it ‘beating’ with the noise from your VFO. It is hard to make a truly low-noise synthesised VFO. Since the direct sampling receiver uses a single fixed sampling frequency, it is much easier to produce a very low noise oscillator and therefore to significantly reduce the effects of reciprocal mixing. The IC-9700 main sampling oscillator is a low-noise module designed originally for sampling audio sound cards. It operates at 49.152MHz before being multiplied to 196.608MHz and sampling the ADC.
On transmit the audio is sampled in an ADC and processed in the FPGA before a 16-bit, dual-channel DAC (Digital to Analogue Converter) IC, sampled at 1179.648MHz, derived from the same source as the ADC clock, is used to produce the 2m and 70cm transmit signals. These are then amplified to the appropriate transmit output power levels.
The 23cm transmit signal is generated as a 311 to 371MHz IF before being converted to the 23cm band in a conventional (transmit) upconverter. An Analogue Devices synthesiser IC is used to generate the 1179MHz sampling clock. A separate LMX2581 synthesiser, with integrated VCO, is used to generate the conversion oscillator frequencies for the 23cm up and down converters.
All modulation and demodulation modes are generated and controlled by the FPGA and an ARM processor.
IC-9700 in Use
My main interest is DXing on 432MHz and up, with some 144MHz FT8 and occasional SSB operating. In order to give the IC-9700 a good ‘airing’ I also used it in satellite mode, made D-STAR and FM voice contacts, as well as listened off the moon on 23cm EME.
My first impression, when listening to beacons on the three bands, was that the signals sounded much cleaner than on my TS2000X, K3 with transverter (2m only), or my FT-847 (2m and 70cm only). This is a common observation by SDR users and is probably because of the low phase noise on the receiver oscillator and lack of distortion that can otherwise occur due to the presence of crystal filters in the superhet receiver path.
After running several test measurements on my IC-9700, I took advantage of the early May 432MHz and Up contest to do a few hours operating during what is normally a very busy contest on both 432 and 1296MHz. However, poor propagation conditions meant there were few 23cm stations to work but 432MHz was busy. I discovered that there was a 144MHz contest on as well and switched to that band on the Sunday in order to find some DX to work and to see how the 144MHz receiver coped with the many more strong signals present. Despite the poor conditions there were many German stations active, together with several Dutch and Belgian stations.
In each case I used my usual antennas, consisting of a 9-element YU7EF on 144MHz, WiMo 23 element Yagi on 432MHz and WiMo 44 element Yagi on 1296MHz. All antennas are horizontally polarised and mounted on my 10m Versatower. A single coaxial cable feed was used on each of the three bands. No masthead preamplifiers or power amplifiers were used, although later I was able to switch in my DG8 144MHz masthead preamplifier in order to test the effectiveness of the external preamplifier facility on the IC-9700.
The IC-9700 transmitter output is quoted by ICOM at 100W on 2m, 75W on 70cm and 10W on 23cm. Table 1 shows the measured power using a 30dB/150W Narda attenuator, 10dB HP attenuator and an HP435 power meter with calibrated HP8481 power meter head.
Table 1: Measured transmitter output. RTTY mode, at 30, 50 and 100% power settings.
Band 30% setting 50% setting 100% setting
144MHz 25W 48W 98W
432MHz 14W 34W 72W
1296MHz 1.9W 4.5W 9W
My impression was that the receiver was quite sensitive on all three bands and this was confirmed my measurements of between 4 and 5dB noise figure, depending on band. This was with the internal preamplifiers (LNAs) turned on. Without the preamplifiers, both 2m and 70cm receivers were exceptionally deaf. The 23cm noise figure, without preamp, was reasonable.
The receiver on 2m and 70cm has a reasonable dynamic range and although the Sherwood receiver comparison tables (URL below) show a relatively poor dynamic range this has so far not been borne out in my on-air testing. The superhet conversion used in the 23cm section, ahead of the direct sampling section of the SDR, works well although the dynamic range is slightly reduced compared with the receivers in the other two bands. Unlike some other 23cm receivers, while beaming directly at it with my 44 element 23cm Yagi, I was able to tune within a couple of kilohertz of the Martlesham GB3MHZ 23cm beacon before the receiver started to respond audibly to the very strong signal. The spectrum display was already showing some ‘distress’ at this spacing. This is regarded as a very good result. The beacon signal level was −53dBm at this point as measured on the S meter.
I asked several of the not-so-busy contest stations for comments on the quality of the transmitted signal on 2m and 70cm. These reports were good to excellent. I used the hand microphone supplied with the radio. Recent measurements by several experienced RF engineers have indicated that the transmitter composite noise output (phase, amplitude and IMD noise) is perhaps not as good as we might have expected from the SDR architecture. More testing is required to confirm the measurements, together with on-air experience.
Testing the transceiver on narrowband FM I was unable to find any local stations to work, so I switched to FM repeater mode. This was easy to set up. The default repeater shift in the UK model on 2m is already −600kHz so I only needed to find the CTCSS tone encode and I was able to access the local Ipswich 2m repeater, GB3PO. I had chosen to test just as the Leiston Radio Society club net was beginning, so was able to solicit audio reports from several club members. They all reported good audio quality. Holding in the ‘XFC’ transmit frequency button I was able to check each signal on the repeater input frequency. Everyone was quite audible, although at slightly lower signal strength than the repeater signal.
I was able to persuade Tony G0MBA to appear on the Clacton-on-Sea 145MHz D-STAR voice repeater, GB7TE C. We were able to chat over the not-too-long 20km path from my QTH to the repeater, with Tony located near to the Clacton repeater. He reported the signal sounded fine, as you would expect with digital voice. We were joined by M0DYS, in Harwich, who also reported a good signal. Both stations were perfectly readable on the repeater input by pressing the XFC button on the IC-9700. From this we decided to change over to direct DV simplex on 145.350MHz. Again, the signal held up well at an indicated level of S5 on the S-meter. It was, however, necessary to use 50W for the QSO. This was with cross polarisation. I used my 9-element horizontally polarized Yagi and G0MBA was using a vertical multiband antenna from a ‘poor location’. Thanks to both Tony and Dillon for the test QSOs.
In order to test the satellite mode, I used my Es’Hail-2 satellite equipment with the IC-9700 in satellite duplex mode. Signals from the narrowband transponder were received at 10.489GHz and converted to 739MHz in my Octagon LNB on a 1m diameter dish. A homebrew down-converter then meant I could tune the transponder passband on 145MHz on the IC-9700. The transmit side on 432MHz was used as the IF for an SG Labs 2400MHz transverter and then into an add-on power amplifier feeding about 10W into a small flat plate antenna.
I found it easy to set up the two frequencies and ‘normal’ tracking. When SATELLITE mode is selected from the MENU button two frequencies are displayed. These can be any two from the three bands available. The two bands can be swapped over and either set as the main band with the other as the sub band. Normal or reverse tracking can be selected by further screen touches on soft buttons.
In this case 145MHz and 435MHz were the two satellite ‘bands’ required. The mode and filter bandwidth are displayed for each band and can be changed, as required, just by touching the appropriate soft ‘button’ on the touchscreen. By touching the frequency display, for either band, that band is shown underlined in yellow and allows the frequency to be changed independently of the other band.
Once the frequency offset is set, the selected band is deselected, and normal or reverse tracking then follows the main tuning dial.
It was convenient to have the spectrum display on to see what signals were present and then to be able to tune to a specific signal and call the station without having to ‘sweep’ around the frequency in order to find your uplink signal and then net onto the called station. Again, reports were favourable with no adverse comments about signal quality.
I used FT8 on 2m to make contacts across nearer Europe under relatively flat band conditions. I was particularly concerned about the frequency stability of the transmitter in narrowband modes such as FT8. Reports have suggested, and confirmed in my own measurements, that there is an apparent flaw in the thermal design around the main clock oscillator, that can lead to a small but noticeable change in the transmission frequency from the time the fan initially switches on during transmit. On 144.174MHz this amounts to a few Hertz. On the higher bands, especially 23cm, the change is significantly larger. The concern is that this change may cause the first few, critical, symbols of the digital transmission to fail to decode correctly at the distant receiver and then again at the local receiver. As the contact progresses the frequency gradually traces back close to the initial value. On 144MHz the effect appears minimal and I was able to enjoy many FT8 contacts. On 432MHz the effect was more pronounced and may have caused several weaker signals not to initially decode. On 23cm it is unlikely that a slow, narrow digital mode such as FT8 would be used very often as the effects of path dispersion would make decoding difficult much of the time, anyway. Wider, fast, modes such as JT9F should be fine because the absolute frequencies are much less critical to successfully decode.
Since the IC-9700 has an external reference frequency input (10MHz at −10dBm) it would not be unreasonable to believe that the clock frequencies in the IC-9700 would be locked to this external reference. This is not the case. The external reference input is merely a modern equivalent of the old crystal marker, where a good internal crystal, with markers every 100kHz or so, is used to check the internal VFO offset. In these older radios the offset could be removed by mechanically adjusting the main tuning dial while holding the VFO on the marker frequency. In the IC-9700 the mechanical setting is replaced by an automatic system based on pressing a ‘soft’ button on the radio touchscreen and letting the radio automatically make the setting correction. The problem is that the main master oscillator is then free to drift again, against the radio operating temperature. Although a 0.5ppm specified 49.153MHz TCXO is used, the thermal design originally did not consider the increased airflow when the fan comes on during transmit. In firmware revision 1.06 it appears that the fan speed may now be better (speed?) controlled. It doesn’t look like an easy job to convert the existing external reference input to work continuously or quasi-continuously, as in the Elecraft K3/K3S transceivers. ICOM are aware of the apparent frequency stability weakness and may bring out a further official modification. We will need to wait and see.
To put it into perspective, the small change in frequency will probably not be noticed when working the traditional CW, SSB, AM and FM modes and most unlikely to be noticed when using the radio’s excellent D-STAR mode. But it is an annoying weakness in the design for use with some narrow digital modes. If it is noticed, it will most likely be on the two higher bands.
MGM modes on the IC-9700
One of the big advantages of the IC-9700, and its HF cousin the IC-7300, is that it can be computer controlled over a single USB connection rather than the usual collection of soundcard interfaces and connecting leads. Provision is still made for the ICOM CI-V interface if required, but the USB connection is both quicker, accommodating up to 115kb/s baud rate, and very comprehensive. It does require the user to download the ICOM USB driver software. Initially plugging in the USB lead into a PC USB socket could result in the installation of Microsoft’s own drivers and consequent loss of some functions. With the USB connection the radio can be both computer-controlled and act as a USB sound card for use with digital modes such as those in the WSJT-X suite of programs.
I set up my IC-9700 as an IC-7300 since WSJT-X version 2.0.1 didn’t recognise the IC-9700. By the time you read this it’s likely that the necessary links for the IC-9700 will have been provided by the main computer control program writers.
With the correct COM port and baud rate selected, to match what I had set on the IC-9700, I was immediately able to use the radio on 144.174MHz, FT8 mode. It did require a bit of adjustment of the sound card audio drive levels so that the radio was just driven to the commencement of ALC. In this mode I was able to work many stations and didn’t observe any frequency drift when reverting to receive between 15 second transmission overs. Most FT8 contacts were made at the 20W level but for a few I turned up the power to around 50W to see if the frequency stability was affected. No decode problems were observed.
The IC-9700 is a very VHF and UHF capable transceiver. In this short review it is impossible to do justice to all its features.
It does have a few niggling omissions such as a single ‘send’ output pin shared by all three bands. If you have three separate linear amplifiers connected, one for each of the three bands, all three will key up at the same time even though only one is wanted. This is a serious error and looks like an oversight by the ICOM designers. All previous multiband transceivers from ICOM have some form of band decode, in the form of a separate send pin on the accessory socket.
For some reason the audio output has not been connected to the centre pin of the microphone socket, even though some headset adapters expect to receive their audio output at this point. The socket wiring diagram shows the connection made but it has been omitted from the production radios.
It is probably not too important for most operators, but I noted that the IC-9700 does not have XIT (transmit incremental tuning). This is particularly useful for me for EME operation when the transmitted signal needs to be placed onto the Doppler frequency shifted echo from a calling station. RIT (receiver incremental tuning) is still provided.
The reference frequency input has already been mentioned.
The sensitivity of the radio is more than adequate for normal use, and the provision to turn off the internal preamplifier and divert DC to power an external preamplifier is a nice touch that overcomes the temptation to ‘pile on the preamplifiers’ with disastrous results for receiver dynamic range.
All in all, while serious weak-signal DXers will probably still opt for a top-of-the-range HF transceiver and pricey VHF/UHF transverters to get the level of performance they demand, the IC-9700 is a lot of radio for the money (bands, modes, features) and, hopefully, will bring more activity to the three bands it covers, especially 23cm. Being software-defined, it should also be future-proofed to some extent – there will undoubtedly be firmware updates as time goes on, in response to user feedback. ICOM should have a big success on their hands. It complements the IC-7300 extremely well, providing coverage of all the UK bands from 472kHz to 70MHz in the IC-7300 and then 144, 432 and 1296MHz in this new radio.
The IC-9700 currently retails for around £1795 from all the major UK dealers. The full specification can be found on the ICOM UK website, along with news of new firmware releases, downloads of the brochure and manual, a video overview and details of complementary accessories. The images in this article are courtesy of ICOM UK.
I would like to thank G7OCD, G8ONH and PA5Y for discussions, measurement assistance and general help with the IC-9700’s SDR architecture. I’d especially like to thank my friend VK7MO for sharing his findings on the radio’s frequency stability issues.
Table 2: IC-9700 receiver noise figures, with and without internal preamplifier.
Band Noise figure, preamp off Noise figure, preamp on
144 17.3dB 4.2dB
432 15.9dB 4.8dB
1296 9.22dB 4.8dB